Characterization of the thymic IL-7 niche in vivoNuno L. Alvesa,b, Odile Richard-Le Goffa,b, Nicholas D. Huntingtona,b, Ana Patricia Sousab,c, Vera S. G. Ribeiroa,b,Allison Bordacka,b, Francina Langa Vivesd, Lucie Pedutoe, Ann Chidgeyf, Ana Cumanob,c, Richard Boydf,Gerard Eberle, and James P. Di Santoa,b,1

aCytokines and Lymphoid Development Unit, cLymphocyte Development Unit, dPlate-Forme Technologique Centre d’Ingenierie Genetique Murine,and eLaboratory of Lymphoid Tissue Development, Institut Pasteur, Paris, France; bINSERM U668, Paris, France; and fMonash Immunology and StemCell Laboratories (MISCL), Monash University, Victoria 3800, Australia

Edited by Max D. Cooper, Emory University, Atlanta, GA, and approved December 2, 2008 (received for review September 25, 2008)

The thymus represents the ‘‘cradle’’ for T cell development, withthymic stroma providing multiple soluble and membrane cues todeveloping thymocytes. Although IL-7 is recognized as an essentialfactor for thymopoiesis, the ‘‘environmental niche’’ of thymic IL-7activity remains poorly characterized in vivo. Using bacterial artificialchromosome transgenic mice in which YFP is under control of IL-7promoter, we identify a subset of thymic epithelial cells (TECs) thatco-express YFP and high levels of Il7 transcripts (IL-7hi cells). IL-7hi TECsarise during early fetal development, persist throughout life, andco-express homeostatic chemokines (Ccl19, Ccl25, Cxcl12) and cyto-kines (Il15) that are critical for normal thymopoiesis. In the adultthymus, IL-7hi cells localize to the cortico-medullary junction anddisplay traits of both cortical and medullary TECs. Interestingly, thefrequency of IL-7hi cells decreases with age, suggesting a mechanismfor the age-related thymic involution that is associated with decliningIL-7 levels. Our temporal-spatial analysis of IL-7-producing cells in thethymus in vivo suggests that thymic IL-7 levels are dynamicallyregulated under distinct physiological conditions. This IL-7 reportermouse provides a valuable tool to further dissect the mechanisms thatgovern thymic IL-7 expression in vivo.

aging � development � thymic epithelial cells � thymus � cytokine

The thymus is the primary organ responsible for the productionand selection of T cells, bearing a diverse T cell receptor

repertoire restricted to self-MHC molecules and tolerant to self-antigens. Anatomically, it is stratified into two main areas: the outercortex and the inner medulla. From fetal through postnatal life,hematopoietic progenitors colonize the thymus and undergo a wellcharacterized differentiation program that occurs in distinct thymicniches. CD4�CD8� double negative (DN) thymocytes committedto the T cell lineage migrate across the cortex, where theyundergo an expansion phase and rearrange the T cell receptor(TCR) � genes. Subsequently, CD4�CD8� double positivethymocytes that successfully rearranged the TCR � genes areselected to proceed to mature single positive (SP) CD4� orCD8� �� T cells, initiating the reverse migratory path across thecortex toward the medulla. As they move through the cortico-medullary junction and into the medulla, they interact withdendritic cells and specialized medullary thymic epithelial cells(mTECs) presenting self-antigens. Auto-reactive T cells arenegatively selected, and mature self-tolerant T cells then exit intothe periphery (1, 2).

The T cell developmental pathway is not autonomous and relieson an organized tri-dimensional cellular network of thymic stromalcells composed of dendritic cells (DCs), macrophages, endothelialcells, fibroblasts, and two specialized subsets of TECs: corticalTECs (cTECs) and mTECs (3). Thymic stroma provides a uniqueplatform of growth factors, chemokines, and receptor ligands thatsustain thymocyte survival, proliferation, migration, and differen-tiation. Compelling evidence suggests a division of labor among thedistinct stromal components during different stages of T celldifferentiation. Whereas mesenchymal-derived fibroblasts andcTECs are important at early stages of T cell development, mTECsand DCs are critical at later stages governing negative selection andproviding survival signals to mature SP thymocytes (3–5). Hence,

one may envisage that, despite the heterogeneity within the stromalcompartment, the different cell types bestow, in a step-wise fashion,specific environmental cues to developing T cells along theirmigration throughout the thymic stromal cell-made ‘‘avenues’’ andin accordance to their requirements.

IL-7 plays a non-redundant role in T cell development. Micelacking IL-7, IL-7R�, or �c chains have a marked reduction in ��T cells and lack �� T cells (6–8). IL-7 belongs to the commoncytokine receptor � chain (�c) cytokine family and binds thereceptor complex consisting of the IL-7R� and �c chains (9).Despite the invaluable contributions deciphering the biologicaleffects of IL-7 on thymocytes, knowledge of the identity ofIL-7-producing cells in vivo, the nature of the distinct anatomicalniches in which IL-7 is active, and the regulatory mechanismsthat govern its expression remain unclear. In vitro and ex vivostudies have shown that IL-7 transcripts can be detected inmultiple cell types and tissues (9–12). Yet, the heterogeneity inthe IL-7 expression between and within the different cellularsubsets hampers the identification of bona fide IL-7-producingcells in vivo. It is conceptually accepted that IL-7 is producedconstitutively under steady-state conditions and exists with alimited availability in the body (9, 13). In parallel, it has beenimplied that a decrease in IL-7 levels is associated with theage-dependent decline in the production of new T cells,generally known has thymic involution (14, 15). Thus, a criticalunresolved issue in lymphocyte homeostasis is how IL-7 isregulated in vivo.

In this study, we generated a novel reporter mouse that monitorsIL-7 expression in vivo. We identified cells co-expressing YFP andhigh levels of Il7 transcripts (IL-7hi cells) that constitute a subset ofTECs detected during early fetal life and that expressed (in additionto IL-7) other key cytokines and chemokines for T cell develop-ment. In the adult thymus, IL-7hi TECs are positioned at thecortico-medullary junction and bear traits of cortical or medullaryTECs. In addition, we provide evidence that IL-7 expression fromTECs declines with age, suggesting that IL-7 levels are under tightcontrol in vivo.

ResultsGeneration of the IL-7 Reporter Mice. To identify IL-7-expressingcells in vivo, we generated transgenic mice carrying a bacterialartificial chromosome (BAC) encoding the YFP under thecontrol of the IL-7 promoter. The YFP gene was inserted byhomologous recombination (16) downstream of the ATG trans-

Author contributions: N.L.A. and J.P.D. designed research; N.L.A., O.R.-L.G., N.D.H., A.P.S.,V.S.G.R., A.B., F.L.V., L.P., A. Chidgey, and A. Cumano performed research; R.B. and G.E.contributed new reagents/analytic tools; N.L.A., N.D.H., and A. Chidgey analyzed data; andN.L.A. and J.P.D. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

1To whom correspondence should be addressed. E-mail: disanto@pasteur.fr.

This article contains supporting information online at www.pnas.org/cgi/content/full/0809559106/DCSupplemental.

© 2009 by The National Academy of Sciences of the USA

1512–1517 � PNAS � February 3, 2009 � vol. 106 � no. 5 www.pnas.org�cgi�doi�10.1073�pnas.0809559106

Dow

nloa

ded

by g

uest

on

Janu

ary

24, 2

020

lational start codon of exon 1 of the IL-7 locus. The transla-tional stop codon of the YFP coding sequence was retained,thus inactivating the BAC IL-7 protein coding sequence on theBAC. The BAC.IL-7.YFP construct was microinjected intoone-cell stage embryos. Five pups had integrated the transgene(screened by PCR analysis of tail DNA), with three showinggerm-line transmission of the transgene. YFP expression wasscreened by immunohistochemistry in fetal and adult thymus,and YFP� cells were detected in progeny of one founder (no.18201) that is further described in this report. BAC.IL-7.YFPmice appeared healthy, and histological examination of 4- to8-week-old mice showed normal development of all organs(data not shown). The cellular composition in distinct primaryand secondary lymphoid organs showed no significant differ-ences compared with age-matched control mice [supportinginformation (SI) Fig. S1].

IL-7hi Cells in the Fetal Thymus Comprise a Subset of Non-Hematopoietic Stromal Cells that Lack Classical Fibroblast and Endo-thelial Markers. Colonization of the thymus by hematopoieticprecursors and subsequent thymopoeisis is initiated during fetal life

(17). We started by examining YFP(IL-7) expression in the thymusat various gestational stages by using YFP-specific antibodies.YFP(IL-7)-expressing cells (which, for simplicity, we will refer to asIL-7hi cells) were detected at E14.5-E17.5 with a scattered distri-bution throughout the thymus. IL-7hi cells were CD45�, indicatinga non-hematopoietic origin (Fig. 1A). We subsequently character-ized the nature of this subset by using a panel of markers todistinguish vascular endothelium (CD31), fibroblasts (ERTR7), orTECs (MHC class II) (11, 18). IL-7hi cells lacked markers ofendothelial cells (CD31) or fibroblasts (ERTR7), but partiallyexpressed MHC class II (Fig. 1B). The phenotypic traits of IL-7hi

cells were maintained at E17.5 (Fig. S2). As TECs are the only cellsthat express MHC class II within the thymic CD45� non-hematopoietic stromal compartment (DCs and macrophages areMHC class II� cells of hematopoietic origin) (3), our data indicatethat IL-7hi cells likely comprise a subset of TECs.

Fetal Thymic IL-7hi Cells Display Traits of TECs. IL-7hi cells were easilydetected by flow cytometry and were larger than CD45� hemato-poietic cells, and their frequency declined concomitantly withthymic development (Fig. 2A). IL-7hi cells were further character-

YFP CD45

YFP ERTR7 CD45

YFP CD31 CD45

YFP MHC II CD45

Tg d14.5

YFP DAPI

Tg d14.5WT d14.5 Tg d17.5A

B

YFP CD45YFP Dapi

Tg d14.5Tg d14.5 Tg d17.5

100 µm

200 µm

WT d14.5

200 µm

100 µm100 µm

200 µm 100 µm

100 µm200 µm200 µm

200 µm

Tg d14.5

Tg d14.5WT d14.5 Tg d14.5

100 µm200 µm200 µm

WT d14.5 Tg d14.5 Tg d14.5

200 µm

Fig. 1. Thymic IL-7hi cells are of non-hematopoieticorigin and lack classical fibroblast and endothelialmarkers. Phenotypic characterization of IL-7hi cells infetal thymus. (A) Immunohistochemical analysis ofBAC.IL-7.YFP transgenic (Tg) and control (WT) fetalthymus at E14.5 and E17.5. (B) Phenotypic character-ization at E14.5 for WT and Tg. Sections were double ortriple stained with the indicated Abs. YFP, green; DAPIand CD45, blue; ERTR7, CD31, and MHC class II, red.White squares indicate IL-7hi cells that co-express MHCclass II. Data are representative of two or threeexperiments.

Alves et al. PNAS � February 3, 2009 � vol. 106 � no. 5 � 1513

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

24, 2

020

ized by using cTEC markers Ly51 and CDR-1 and the mTECmarker lectin Ulex Europaeus Agglutinin (UEA) I (11, 18). AtE13.5, approximately 60% of IL-7hi cells co-expressed MHC classII (data not shown). From E14.5 onward, the majority of IL-7hi cellsco-expressed MHC class II, CDR1, and Ly51. Only a minorproportion of thymic stromal cells bound UEA at E17.5, themajority of which were included in the YFP� fraction (Fig. 2B).Thus, our data are consistent with fetal thymic IL-7hi cells being asubset of TECs.

Il7 mRNA levels in fetal thymic IL-7hi cells were next analyzedby quantitative PCR in sorted MHCII�YFP� and MHCII�YFP�

cells. As shown in Fig. 2C, Il7 transcripts were increased 20-fold inMHCII�YFP� cells compared with their MHCII�YFP� counter-parts. We performed an analysis on the gene expression profile offetal thymic IL-7hi cells for 34 cytokines and 37 chemokines. Otherthan Il7, a subset of cytokines and chemokines were up-regulatedin IL-7hi cells, including the �c cytokine Il15, the homeostaticchemokines Ccl25, Cxcl12, and Ccl19 and the inflammatory che-mokines Cxcl10, Tnfsf15, and Ccl5 (Fig. 2D). Additional analysisconfirmed these observations and revealed that transcripts forNotch ligands, such as Dll4 and Jag2, as well as the transcriptionfactor Foxn1, were up-regulated in IL-7hi cells (data not shown).Together, our data demonstrate that IL-7hi cells co-transcribe anarray of receptor ligands, cytokines, and chemokines involved in thesurvival, attraction, proliferation, and maturation of thymocytes.

In the Adult Thymus, IL-7hi Cells Are Preferentially Localized at theCortico-Medullary Junction and Bear Markers of Both Cortical andMedullary TECs. Thymic epithelial patterning is initiated during fetaldevelopment and continues throughout postnatal life, with the finalacquisition of the typical cortex-medulla compartmentalizationachieved in adult life (3, 19). Given the dynamic nature andcontinual restructuring of thymic epithelium, we assessed whetherIL-7hi cells preserved their lineage origin after birth. IL-7hi cellswere present in postnatal thymus, displayed TEC traits similar totheir fetal thymic counterparts (CD45� MHC class II� Ly51�

CDR-1� UEA�), and clearly expressed abundant Il7 transcripts.TECs that lack YFP also expressed Il7, although at lower levels thanIL-7hi TECs (Fig. 2 E and F and Fig. S3). These data indicate thatthis reporter mouse identifies TECs that express abundant Il7transcripts and suggests the existence of heterogeneity of IL-7expression within TECs.

The spatial location and phenotype of IL-7hi cells in adult thymuscells were characterized. IL-7hi cells were predominantly localizedat the cortico-medullary junction and inside the medulla, definedeither by nuclear staining (with DAPI) or MTS33 staining. Notethat DAPI staining defines high and low cellular density areas of thecortex and medulla, respectively, and MTS33 stains cortical but notmedullary thymocytes and small clusters of medullary epithelialcells (Fig. 3A) (20). Thus, IL-7hi cells are strategically positioned atan anatomical site favorable for interactions with thymocytes atboth early and late stages of their development (5).

Analysis of Il7 mRNA levels in adult thymic stromal subsetsrevealed that cortical and medullary TECs were the two predom-inant IL-7-expressing subsets (Fig. S4A). Accordingly, we foundthat adult thymic IL-7hi cells lacked markers of hematopoietic cells(CD45), endothelial cells (CD31), and fibroblasts (ERTR7;

and Actb). Depicted is the average difference (in fold) between MHCII�YFP�

and MHCII�YFP� cells from two independent experiments. (E,F) Il7 transcriptsare markedly up-regulated in newborn thymic IL-7hi TECs. (E) Analysis wasperformed on thymic stroma cells (CD45� gate). (F) The relative expression ofIl7 mRNA levels was assessed by qPCR (Right) and normalized to Hprt in sortedthymic MHCII�YFP� and MHCII�YFP� cells and compared relative toMHCII�YFP� cells, in which Il7 levels were set to 1. Depicted is the averagedifference (in fold) between the two cellular subsets from three independentexperiments.

0 50K 100K 150K 200K 250K0

50K

100K

150K

200K

250K

B MHC II CDR1 Ly51

A

E14.5

E17.5

C

UEA

YFP (FL1)

CD

45

YFP(FL1)

FSCS

SC

N.D.

0 102 103 104 105

0

102

103

104

105

43.5

14.7

E14.5

0 102 103 104 105

0

102

103

104

105

0.23

37.519.7

0 102 103 104 105

0

102

103

104

105

19.2 40

2.59

0 102 103 104 105

0

102

103

104

105

40.8

1.14

16

0 102 103 104 105

0

102

103

104

105

26.320.9

1.67

0 102 103 104 105

0

102

103

104

105

25.3

17.7

0.49

0 102 103 104 105

0

102

103

104

105

3.43 0.75

26.3

0 102 103 104 105

0

102

103

104

105

23.1

5.75

17.1

E14.5 E17.5E13.5M

HC

II- Y

FP

-

MH

CII+

YF

P+

Fol

ddi

ffere

nces

rela

tive

mR

NA

leve

ls

Il7 (E14.5)

D

Il7 Il15

Ccl25

Cxcl12

Ccl19

Ccl21

Cx3cl1

Ccl17

Cxcl10

Tnfsf15

Ccl5

Cxcl9

Cxcl11

Ccl3

Homeostatic

cytokines

Homeostatic

chemokines

Inflammatory

chemokines

0

5

10

15

20

25

30

Fol

ddi

ffere

nce

rela

tive

mR

NA

leve

ls

0 50K 100K 150K 200K 250K0

50K

100K

150K

200K

250K

0 50K 100K 150K 200K 250KFSC A

0

50K

100K

150K

200K

250K

14.743.5

1.6794.7

22.929.7

Fol

ddi

ffere

nces

rela

tive

mR

NA

leve

ls

E Neonatal

0 102 103 104 105

0

102

103

104

105

8.11 8.08

79.8

YFP(FL1)

MH

CII

0

100

200

300

400

500

YFP-

0

10

20

30

YFP+

TECs

Il7(neonatal)F

Fig. 2. IL-7hi cells display traits of TECs and transcribe homeostatic cytokines andchemokines important for thymocyte development. Flow cytometric analysis offetal (E13.5, E14.5, and E17.5) and neonatal thymic IL-7hi cells from BAC.IL-7.YFPtransgenic mice (Tg) following stromal isolation. (A) Thymic IL-7hi cells compriseasubsetof stromal (CD45�) cells. Thesizeof IL-7hi cells (green)wascomparedwiththat of CD45� cells (blue). (B) IL-7hi cells were characterized phenotypically for theindicated TEC surface markers. Analysis was performed on thymic stromal cells(CD45� gate). (A,B) Numbers indicate the percentage of each correspondentboxed gate. (N.D., not detectable.) Data are representative of two or threeexperiments. (C,D) Transcriptional analysis was performed in sorted fetal thymicMHCII�YFP� and MHCII�YFP� stromal (CD45�) populations from BAC.IL-7.YFPfetuses (E14.5). (C) Il7 transcripts are markedly up-regulated in fetal thymic IL-7hi

cells. The relative expression of Il7 mRNA levels was assessed by quantitative PCRand normalized to Hprt. Thymic MHCII�YFP� cells were compared relative toMHCII�YFP� cells, in which Il7 levels were set to 1. Depicted is the averagedifference (in fold) between the two cellular subsets from six independentexperiments. (D) The relative mRNA expression for 34 cytokines and 37 chemo-kines was assessed in fetal thymic IL-7hi cells by RT-PCR array. The mRNA level foreach target gene was normalized to three house-keeping genes (Hspcb, Gapdh,

1514 � www.pnas.org�cgi�doi�10.1073�pnas.0809559106 Alves et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

24, 2

020

Fig. 3A). The two major subsets of TECs (cTECs and mTECs) canbe distinguished in the adult thymus on the basis of their anatom-ical, phenotypic, and functional features. Given the location ofIL-7hi cells at the cortico-medullary junction, we further charac-terized them for the expression of cTEC (Ly51 and CDR-1) andmTEC (MTS-10 and UEA) markers. IL-7hi cells located near thecortical area were CDR-1�Ly51�UEA�MTS10�, whereas those

proximal to the medulla were CDR-1�Ly51�MTS10�UEA� (Fig.3B). Together, our results indicate that, in adult thymus, IL-7hi cellscomprise at least two specialized subsets bearing traits of eithercortical or medullary TECs.

YFP�(IL-7hi) Cells Are Not Detected in the Periphery of BAC.IL-7.YFPTransgenic Mice. Apart from its role in thymocyte development,IL-7 is also critical for the control of peripheral T cell homeostasisand B cell development (7, 8, 21). We therefore examined BAC.IL-7.YFP reporter mice for IL-7hi cells outside of the thymus. Histo-logical and flow cytometric analyses of 3- to 8-week-old miceshowed that YFP� cells were not detected in the bone marrow andsecondary lymphoid organs, including spleen and peripheral lymphnodes (data not shown). A probable explanation for this result isthat our reporter mouse may not faithfully read-out the completerange of IL-7 expression in vivo. Previous studies have identified Il7transcripts in several cell types, including lymph node reticular cells,thymic DC cells, and fibroblast and gut epithelial cells (9–12). Wehypothesized that these cells might have lower IL-7 expressioncompared with thymic IL-7hi TECs that we are able to visualize inBAC.IL-7.YFP transgenic mice. The glycoprotein gp38 defines apopulation of IL-7-expressing fibroblast reticular cells in the lymphnodes (10). We therefore compared Il7 mRNA expression levelsfrom sorted thymic IL-7� TECs and gp38� fibroblastic lymph nodereticular cells. We found that Il7 transcripts were more stronglyexpressed in IL-7hi TECs compared with gp38� fibroblasts (Fig.S4B). Thus, it is likely that our reporter mouse can clearly identifyonly cells expressing high IL-7 levels.

The Frequency of Thymic IL-7hi Cells Declines with Age. Thymicinvolution has been associated with a decline in IL-7 levels (14, 15).These changes appear to affect and/or be affected by the cellularcomposition of the thymic stroma. To address whether the age-dependent decline in T cell production was associated with alter-ations in IL-7hi TECs, we analyzed YFP expression in TECs of miceat different ages. As shown in Fig. 4A, and in accordance withprevious studies (18), the number of CD45� cells reached theirmaximum between 2 and 7 weeks, followed by a conspicuousdecline. Although both TECs subsets—MHCII�YFP� andMHCII�YFP�—showed a similar pattern of age-related loss, thereduction in IL-7hi cells was more profound (Fig. 4 B–D). Thesedata provide in vivo evidence that the decline in T cell productionwith age may be a consequence of a diminution in TECs thatexpress high levels of IL-7.

DiscussionDespite the accumulating knowledge on the biological effects ofIL-7 during thymopoiesis, the precise identity of IL-7-expressingcells in vivo has eluded immunologists. It has been proposed thatthymic stromal components are the source of IL-7 (19, 22). Yet,the lack of means to specifically identify IL-7-expressing cellsprecludes the resolution of important issues in fundamentalimmunology, such as their developmental lineage, anatomicallocation, and associated functional characteristics. Here, wedescribe an IL-7 reporter mouse that allowed us to identify andcharacterize TECs expressing high levels of IL-7. Our dataprovide evidence that IL-7hi cells are confined to a TEC subsetdetected during early fetal development and that persistthroughout postnatal life. In the adult thymus, IL-7hi cells arepredominantly localized at the cortico-medullary junction anddisplay features of cortical or medullary TECs.

The thymus starts to be colonized by hematopoietic progenitorsby days 11 to 12 of gestation (23). Over the next 2 days, IL-7-dependent T cell precursors (DN2 and DN3, defined asCD44�CD25� and CD44�CD25�) are first observed (24), suggest-ing that IL-7 expression is initiated during this period. Interestingly,in association with the presumable need for IL-7, IL-7hi cells weredetected within this time frame (after E13.5; Figs. 1 and 2). Our

YFP MTS10 DAPI

YFP DAPIA

500 µm

C

M

YFP CD31 CD45 YFP ERTR7CD45

100 µm 100 µm

B

YFP MTS33 CD45

200 µm

YFP UEA Ly51

C

M

100 µm

YFP CDR1 DAPI

C

M

100 µm

C

M

100 µm

Fig. 3. Adult thymic IL-7hi cells are positioned at cortico-medullary regionsand bear markers of cortical and medullary TECs. Immunohistochemical anal-ysis of adult thymus (6–8 weeks) of BAC.IL-7.YFP transgenic mice. Sectionswere double or triple stained with the indicated Ab combination (C, cortex; M,medulla). (A) IL-7hi cells are of nonhematopoietic origin, and predominantlypositioned at the cortico-medullary junction and do not costain with endo-thelial or fibroblast markers [YFP, green; DAPI and CD45 blue; MTS33 (ThB),ERTR7, CD31, red]. (B) IL-7hi cells display markers of cTECs or mTECs. Arrowsindicate IL-7hi cells that either coexpress cTEC (CDR1 and Ly51) or mTEC(MTS10) markers (eYFP, green; DAPI, Ly51, blue; CDR1, UEA, MTS10, red). Dataare representative of 2–3 experiments.

Alves et al. PNAS � February 3, 2009 � vol. 106 � no. 5 � 1515

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

24, 2

020

data strongly suggest that this subset comprise a population ofIL-7-expressing TECs. First, Il7 transcripts were markedly up-regulated in YFP� cells (Fig. 2). Second, IL-7hi cells were ofnon-hematopoietic origin (CD45�), lacked endothelial and mes-enchymal markers, and expressed TEC markers throughout devel-opment, including MHC class II, Ly51, and CDR1 (Figs. 1 and 2).Last, transcripts for Foxn1, a transcription factor crucial for TECdifferentiation (25), were up-regulated in IL-7hi cells, corroboratinga TEC signature (data not shown). Interestingly, various transcriptsfor homeostatic cytokines, chemokines, and receptor ligands thatare critical for T cell development were found to be co-expressedin IL-7hi cells (Fig. 2 and data not shown). These included thechemokines CCL25 (thymus-expressed chemokine) and CXCL12(stromal cell-derived factor-1), which are important for the recruit-ment of hematopoietic progenitors into the thymus, and the ligandsfor FMS-like tyrosine kinase 3 receptor and Notch receptors, whichare important for the survival and commitment of early T cellprecursors (26). Based on these results, it is possible that IL-7hi

TECs may play a broader role at early stages of thymopoesis.The complete patterning of the thymus, typified by the segre-

gation into cortex and medulla, is a dynamic process initiated duringfetal life and ultimately achieved in adulthood (19). In the fetalthymus, IL-7hi cells were found scattered in the fetal thymus.Interestingly, however, in the adult thymus this subset was strate-gically positioned at the cortico-medullary region, bearing traits ofcortical (Ly51�CDR-1�MTS10�) or medullary (Ly51�CDR-1�MTS10�) TECs (Fig. 3). Concordantly, in the adult thymus, Il7transcripts were predominantly expressed in fractionated corticaland medullary TECs (Fig. S4A). These observations suggest theexistence of two IL-7hi TEC subsets within the adult thymus.Compelling evidence has shed light on the lineage relationshipsbetween cortical and medullary TECs. It is currently accepted thatthese two subsets are derived from a common epithelial progenitor

(27–29). Given the anatomical location and the exclusive expressionof either cortical or medullary markers, it remains to be addressedwhether cortical and medullary IL-7hi cells constitute intermediateprogenitor subsets of mature cTECs and mTECs, respectively.

Conceptual models propose that distinct stromal componentsplay specialized roles nurturing thymocytes, in a step-wise fashion,with various environmental cues that sustain their commitment(Notch ligands), survival (IL-7), migration (thymus-expressed che-mokine, stromal cell-derived factor-1, CCL19, and CCL21), andselection (promiscuous gene expression by AIRE� mTECs) (2, 3,5, 19). It is interesting to note that the anatomical location of cellsexpressing high levels of IL-7 is strikingly associated with therequirements for IL-7 by thymocytes at different stages of differ-entiation. IL-7-derived survival and proliferative signals are neededby early T cell precursors positioned in inner part of the cortex forthe DN2-DN3 transition and by SP mature T cells located in themedulla (6, 30). In this regard, we postulate a division of laboramong IL-7hi cells, such that the IL-7hiCDR1�LY51�MTS10� andIL-7hiCDR1�Ly51�MTS10� TECs are responsible for nurturingthymocytes at early and late stages of development, respectively.

Thymic stromal lymphopoietin (TSLP), another IL-7-like cyto-kine (31), appears to be expressed by human epithelial cells ofHassall corpuscle in the thymic medulla (32). Despite their closebiological activities (33), we found that TSLP was not enriched inIL-7hi cells (data not shown), suggesting that IL-7 and TSLP nichesmay be cellular and anatomical distinct. Both IL-7 and TSLP playalso an important role in B cell development and peripheral T cellhomeostasis (7, 8, 21, 33). Yet, a comprehensive analysis of normalBAC.IL-7.YFP transgenic mice in multiple lymphoid organs has sofar unsuccessfully revealed any YFP� cells in anatomical locationsother than the thymus. It is likely that this novel reporter mouse mayidentify only cells expressing high levels of IL-7 (Fig. 2 and Fig. S4),which explains the lack of YFP expression in other organs. Thisobservation may be associated with distinct physiological context,which reflects a putative lower triggering threshold for IL-7 inperipheral mature T cells.

Thymic involution is associated with reduction in T cell output.It has been proposed that diminishing IL-7 level is associated withthe decrease in T cell generation, although experimental evidenceis limited (14, 15). Our results suggest there is heterogeneity in IL-7expression within TECs (Fig. 2E) and alterations in the dynamic ofIL-7-expressing TEC populations with time. Particularly, the num-ber of thymic IL-7hi cells diminishes with age, and remarkably, thisbegins as early as 5 to 7 weeks of age. Although an overall decreasein the stromal cellularity was noted, in line with previous observa-tions (18), we found that the frequency of YFP�(IL-7hi) TECsdeclined at a faster rate compared than YFP� TECs (Fig. 4). Itremains to be answered whether the reduction in IL-7hi TECs is acause or consequence of thymic involution, and whether theage-associated decrrease in their density is reflected by a decreasein the functional capacity to transcribe Il7 or by impaired turnoverof this specialized subset.

In conclusion, our data provide a precise anatomical and phe-notypical characterization of IL-7-expressing TECs in the thymus invivo. The identification of a thymic IL-7-dependent ‘‘niche’’ willprovide the opportunity to address questions concerning the inter-action between different lymphocyte populations with IL-7-producing cells in normal and pathological settings. This reportermouse provides an unique tool to survey the major IL-7-producingcells. Moreover, further transcriptional analysis of cells expressinghigh levels of IL-7 should provide a framework to gain insights intothe molecular mechanisms that regulate its expression. Under-standing the rules that govern IL-7 expression will have importantimplications for therapies that aim to boost T cell developmentand/or homeostasis through increased IL-7 availability.

106

107

108

109An

um

be

ro

fC

D4

5+

ce

lls

iso

late

d/t

hym

us

neonates 6 months5-7 weeks

4.45 2.8

2.06 0.46

D

10.66.95

MH

CII

neonat

es

1-2 wks

5-7 wks

3-6m

onths

+6 month

s103

104

105

106B

nu

mb

er

of

ce

lls

iso

late

d/t

hym

us

neonat

es

1-2 wks

5-7 wks

3-6 m

onths

+6 month

s

es0

4

8

12

16C

%w

ith

inC

D4

5-g

ate

neonat

es

1-2 wks

5-7 wks

3-6 m

onths

+6 month

s

*

*

**

MHCII+YFP+

MHCII+YFP-

YFP(FL1)

MHCII+YFP+

MHCII+YFP-

Fig. 4. The frequency of thymic IL-7hi TECs declines with age. Flow cytometricanalysis of IL-7hi TECs isolated from BAC.IL-7.YFP transgenic thymus at differentages. (A) Number of CD45� cells. (B and C) Number and percentage of stromal(CD45�) MHCII�YFP� cells (gray bars) and MHCII�YFP� cells (black bars). (For alltimepoints,n�7–9,except�6months,n�3.)Differencesweresignificantwhen2-week and 3- to 6-month groups were compared (CD45� and MHCII�YFP�, P �0.0003; MHCII�YFP�, P � 0.0037). (D) Flow cytometric analysis showing a repre-sentative young, adult, and old thymus (gating on CD45� cells). Numbers indicatethe percentage of the correspondent boxed gates.

1516 � www.pnas.org�cgi�doi�10.1073�pnas.0809559106 Alves et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

24, 2

020

Materials and MethodsGeneration of BAC.IL-7.YFP Transgenic Mice. Mice with a YFP reporter geneunder the control of the IL-7 regulatory sequences were generated in the SJLbackground essentially as described previously (16), and backcrossed for 4 to 5generations to C57BL/6 mice. For thymic stromal cell subset isolation, C57BL/6Jmice aged 6 to 7 weeks were obtained from Monash Animal Services. All micewere kept in specific pathogen-free conditions, and all animal experiments wereapproved by the committee on animal experimentation at Institut Pasteur (Paris,France) and were approved by the French Ministry of Agriculture.

Isolation of Thymic Stromal Cells. Thymic stromal cells were isolated as described(34). Thymic fragments were digested for 30 min at 37 °C in RPMI medium 1640containing 0.125% (wt/vol) collagenase D (Roche Molecular Biochemicals) and0.1%(wt/vol)DNase I (Roche), followedby0.1%(wt/vol) trypsin(RocheMolecularBiochemicals) and 0.1% (wt/vol) DNase I.

Antibodies. Phycoerythrin (PE)-conjugated anti-Ter-119 (Ly-76), PE-conjugatedanti-CD4 (RM4–5), allophycocyanin-cyanine 7 (APCCy7)-conjugated anti-CD19(1D3), allophycocyanin (APC)-conjugated anti-B220 (RA3–6B2), APCCy7-conjugated anti-CD11b (M1/70), peridinin chlorophyll protein-cyanine 5.5 (Per-CPCy5.5)-conjugated anti-CD45.2 (104), Alexa 750-conjugated anti-CD44 (IM7),and biotin-conjugated anti-Ly51(6C3) were purchased from Becton DickinsonPharMingen. PE-conjugated anti-I-A/I-E, PerCPCy5.5-conjugated anti-CD3 (145–2C11), phycoerythrin-cyanine 7(PECy7)-conjugated anti-GR-1 (Ly-6G RB6.8C5),APCCy7-conjugated anti-CD45.2 (104), PECy7-conjugated anti-CD25 (PC61), andbiotin-conjugated anti-CD45.2 (104) were purchased from eBiosciences. PE-conjugated anti-IgM and APC-conjugated anti-CD8 (Ly-2) were purchased fromSouthern Biotechnology Associates. Purified anti-ERTR7 was purchased fromBMA Biomedicals. Purified anti-GFP (A-11122), Alexa 488-conjugated anti-rabbit,Alexa 555-conjugated anti-rat, and Alexa 647- or Alexa 555-conjugated strepta-vidins were purchased form Invitrogen. Cy3-anti-syrian hamster was purchasedfromJacksonImmunoresearch, rhodamine-conjugatedUEA1wasobtainedfromVectorlabs, and anti-gp38 culture supernatant was a gift from A. Farr (Seattle,WA). The anti-CD31 (MTS-12), anti-CDR-1, and anti-MTS-10 antibodies have beendescribed previously (18).

Flow Cytometry. Cells isolated from thymus, lymph node, spleen, and bonemarrow were prepared as described (35). Before staining, cells were treated with

Fc-Block (antibody to CD16-CD32; Becton Dickinson PharMingen). Analysis wasdone with an FACSCanto (Becton Dickinson) and FlowJo software (Treestar). Cellsorting was performed by using FACSAria (BD Biosciences).

Immunostaining. Thymi were prepared as described previously (36). Briefly,organs were fixed in 4% paraformaldehyde (Sigma) in PBS solution, washed inPBS solution, incubated in a solution of 30% sucrose (Sigma), embedded in OCTcompound (Sakura), and frozen. Frozen sections (8 �m) were cut with an OMcryostat (HM500; Microm) and collected onto Superfrost/Plus slides (Fisher Scien-tific). After blocking (10% BSA in PBS solution), samples were incubated withprimary antibodies followed by secondary reagents, with DAPI (Sigma), andmounted with Fluoromount-G (Southern Biotechnology Associates). Slides wereexamined under an AxioImager M1 fluorescence microscope (Zeiss) equippedwith a CCD camera, and images were processed with AxioVision software (Zeiss).

Gene Expression. For quantitative PCR (qPCR), mRNA was purified with anRNease Mini Kit (Qiagen). The quality of total RNA was assessed by using the 2100Bioanalyzer system (Agilent Technologies). RNA was reverse transcribed usingSuperScript III first-strand synthesis system for RT-PCR (Invitrogen) and oligo(dT)oligonucleotides according to the manufacturer’s instructions. The cDNA wassubjected to qPCR using TaqMan Universal Master Mix (Applied Biosystems) orPlatinum SYBR Green qPCR Supermix UDG (Invitrogen). Detection was per-formed using an ABI Prism 7000 Sequence Detection System (Applied Biosystems)or a Corbett Rotor-Gene 3000 (Corbett Research). Primers for Hprt and Il-7 werefrom Applied Biosystems; and for Gapdh and Il-7 were from Qiagen. The delta-delta-Ct method was used to calculate relative levels of target mRNA comparedwith Gapdh or Hprt.

For the gene expression profile, 500 ng of total RNA was converted to cDNAby using RT2 PCR array first-strand kit (SuperArray Bioscience). All procedureswere performed according to the manufacturer’s protocols. The expression of 84different cytokine and chemokine genes was measured using the custom-mademouse RT2 Profiler PCR array (SuperArray Bioscience). Real-time PCR was per-formed on a PTC-200 thermocycler equipped with a Chromo4 detector (Bio-Rad).Data were analyzed using Opticon Monitor software (Bio-Rad).

ACKNOWLEDGMENTS. We thank S. Lesjean-Pottier for technical support. Wethank Dr. Paulo Vieira for helpful discussions and critical reading of the manu-script. This work was funded by Institut Pasteur, INSERM, and the Ligue NationaleContre leCancer (EquipeLabellisee).N.L.A.was fundedbya long-termfellowshipfrom the European Molecular Biology Organization.

1. Jameson SC, Hogquist KA, Bevan MJ (1995) Positive selection of thymocytes. Annu RevImmunol 13:93–126.

2. Kyewski B, Klein L (2006) A central role for central tolerance. Annu Rev Immunol24:571–606.

3. Anderson G, Moore NC, Owen JJ, Jenkinson EJ (1996) Cellular interactions in thymocytedevelopment. Annu Rev Immunol 14:73–99.

4. Derbinski J, Schulte A, Kyewski B, Klein L (2001) Promiscuous gene expression inmedullary thymic epithelial cells mirrors the peripheral self. Nat Immunol 2:1032–1039.

5. Petrie HT, Zuniga-Pflucker JC (2007) Zoned out: functional mapping of stromal signal-ing microenvironments in the thymus. Annu Rev Immunol 25:649–679.

6. DiSanto JP, Muller W, Guy-Grand D, Fischer A, Rajewsky K (1995) Lymphoid develop-ment in mice with a targeted deletion of the interleukin 2 receptor gamma chain. ProcNatl Acad Sci USA 92:377–381.

7. von Freeden-Jeffry U, et al. (1995) Lymphopenia in interleukin (IL)-7 gene-deleted miceidentifies IL-7 as a nonredundant cytokine. J Exp Med 181:1519–1526.

8. Peschon JJ, et al. (1994) Early lymphocyte expansion is severely impaired in interleukin7 receptor-deficient mice. J Exp Med 180:1955–1960.

9. Fry TJ, Mackall CL (2002) Interleukin-7: from bench to clinic. Blood 99:3892–3904.10. Link A, et al. (2007) Fibroblastic reticular cells in lymph nodes regulate the homeostasis

of naive T cells. Nat Immunol 8:1255–1265.11. Gray DH, et al. (2007) A unique thymic fibroblast population revealed by the mono-

clonal antibody MTS-15. J Immunol 178:4956–4965.12. Zamisch M, et al. (2005) Ontogeny and regulation of IL-7-expressing thymic epithelial

cells. J Immunol 174:60–67.13. Napolitano LA, et al. (2001) Increased production of IL-7 accompanies HIV-1-mediated

T-cell depletion: implications for T-cell homeostasis. Nat Med 7:73–79.14. Aspinall R (2006) T cell development, ageing and Interleukin-7. Mech Ageing Dev

127:572–578.15. Chidgey A, Dudakov J, Seach N, Boyd R (2007) Impact of niche aging on thymic

regeneration and immune reconstitution. Semin Immunol 19:331–340.16. Sparwasser T, Gong S, Li JY, Eberl G (2004) General method for the modification of

different BAC types and the rapid generation of BAC transgenic mice. Genesis 38:39–50.

17. Jotereau F, Heuze F, Salomon-Vie V, Gascan H (1987) Cell kinetics in the fetal mousethymus: precursor cell input, proliferation, and emigration. J Immunol 138:1026–1030.

18. Gray DH, et al. (2006) Developmental kinetics, turnover, and stimulatory capacity ofthymic epithelial cells. Blood 108:3777–3785.

19. Anderson G, Lane PJ, Jenkinson EJ (2007) Generating intrathymic microenvironmentsto establish T-cell tolerance. Nat Rev Immunol 7:954–963.

20. Takeoka Y, et al. (1997) A comparative analysis of the murine thymic microenviron-ment in normal, autoimmune, and immunodeficiency states. Dev Immunol 5:79–89.

21. Schluns KS, Kieper WC, Jameson SC, Lefrancois L (2000) Cytokine control of memoryT-cell development and survival. Nat Immunol 1:426–432.

22. Anderson G, Jenkinson EJ (2001) Lymphostromal interactions in thymic developmentand function. Nat Rev Immunol 1:31–40.

23. Owen JJ, Ritter MA (1969) Tissue interaction in the development of thymus lympho-cytes. J Exp Med 129:431–442.

24. Douagi I, Vieira P, Cumano A (2002) Lymphocyte commitment during embryonicdevelopment, in the mouse. Semin Immunol 14:361–369.

25. Nehls M, et al. (1996) Two genetically separable steps in the differentiation of thymicepithelium. Science 272:886–889.

26. Jenkinson EJ, Jenkinson WE, Rossi SW, Anderson G (2006) The thymus and T-cellcommitment: the right niche for Notch? Nat Rev Immunol 6:551–555.

27. Gill J, Malin M, Hollander GA, Boyd R (2002) Generation of a complete thymicmicroenvironment by MTS24(�) thymic epithelial cells. Nat Immunol 3:635–642.

28. Rossi SW, Jenkinson WE, Anderson G, Jenkinson EJ (2006) Clonal analysis reveals acommon progenitor for thymic cortical and medullary epithelium. Nature 441:988–991.

29. Bleul CC, et al. (2006) Formation of a functional thymus initiated by a postnatalepithelial progenitor cell. Nature 441:992–996.

30. Di Santo JP, et al. (1999) The common cytokine receptor gamma chain and the pre-T cellreceptor provide independent but critically overlapping signals in early alpha/beta Tcell development. J Exp Med 189:563–574.

31. Friend SL, et al. (1994) A thymic stromal cell line supports in vitro development ofsurface IgM� B cells and produces a novel growth factor affecting B and T lineage cells.Exp Hematol 22:321–328.

32. Watanabe N, et al. (2005) Hassall’s corpuscles instruct dendritic cells to induceCD4�CD25� regulatory T cells in human thymus. Nature 436:1181–1185.

33. Liu YJ, et al. (2007) TSLP: an epithelial cell cytokine that regulates T cell differentiationby conditioning dendritic cell maturation. Annu Rev Immunol 25:193–219.

34. Gray DH, Chidgey AP, Boyd RL (2002) Analysis of thymic stromal cell populations usingflow cytometry. J Immunol Methods 260:15–28.

35. Samson SI, et al. (2003) GATA-3 promotes maturation, IFN-gamma production, andliver-specific homing of NK cells. Immunity 19:701–711.

36. Eberl G, et al. (2004) An essential function for the nuclear receptor RORgamma(t) in thegeneration of fetal lymphoid tissue inducer cells. Nat Immunol 5:64–73.

Alves et al. PNAS � February 3, 2009 � vol. 106 � no. 5 � 1517

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

24, 2

020