Deletion Upon MHC Class I-Restricted Thymocyte

Home , Clonal deletion

CD11c+ Eosinophils in the Murine Thymus: Developmental Regulation and Recruitment upon MHC Class I-Restricted Thymocyte Deletion This information is current as of September 24, 2021. Mark Throsby, André Herbelin, Jean-Marie Pléau and Mireille Dardenne J Immunol 2000; 165:1965-1975; ; doi: 10.4049/jimmunol.165.4.1965 http://www.jimmunol.org/content/165/4/1965 Downloaded from

References This article cites 48 articles, 22 of which you can access for free at:

http://www.jimmunol.org/content/165/4/1965.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on September 24, 2021

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2000 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. CD11c؉ Eosinophils in the Murine Thymus: Developmental Regulation and Recruitment upon MHC Class I-Restricted Thymocyte Deletion

Mark Throsby,1* Andre´Herbelin,† Jean-Marie Ple´au,* and Mireille Dardenne*

Eosinophils are bone marrow-derived cells released into the circulation during hypersensitivity reactions and parasitic infections. Under normal conditions most eosinophils are tissue bound, where their physiologic role is unclear. During in situ analysis of the thymic microenvironment for CD11c؉ dendritic cell subpopulations (APC critical in the process of thymic negative selection) a discrete population of CD11b/CD11c double-positive cells concentrated in the cortico-medullary region of young mice was detected. Thymic CD11c؉ cells were isolated, and the CD11b؉ subpopulation (CD44high, class IIlow, CD11cint) was identiﬁed as mature eosinophils based on: scatter characteristics, major basic protein mRNA expression, and eosinophilic granules. They are Downloaded from hypodense, release high levels of superoxide anion, and express CD25, CD69, and mRNA for IL-4 and IL-13, but not GM-CSF or IL-5, suggesting a distinct state of activation. Thymic eosinophils are preferentially recruited during the neonatal period; absolute numbers increased 10-fold between 7–14 days to reach parity with dendritic cells before diminishing. In a model of acute negative selection, eosinophil numbers were increased 2-fold 6 h after cognate peptide injection into MHC class I-restricted female H-Y TCR transgenic mice. In both peptide-treated female and negatively selecting male H-Y TCR mice, clusters of apoptotic

bodies were associated with eosinophils throughout the thymus. Our data demonstrate a temporal and spatial association between http://www.jimmunol.org/ eosinophil recruitment and class I-restricted selection in the thymus, suggesting an immunomodulatory role for eosinophils under nonpathological conditions. The Journal of Immunology, 2000, 165: 1965–1975.

he maturation and differentiation of T lymphocytes occur tionally associated with host defense against parasitic infection, primarily within the thymic microenvironment. After through degranulation and release of toxic proteins, eosinophils T maturation, migrating cells are self-restricted and purged also produce a variety of pro- and anti-inflammatory cytokines, of high affinity reactivity with common self-Ags (1, 2). The selec- growth factors, and chemokines (10–13). Eosinophils can act as tion of the final circulating repertoire occurs through direct inter- APC (14, 15) and express several important costimulatory mole- action with epithelial and hemopoietic elements of the thymic mi- cules (11, 16), indicating that they could modulate adaptive im- by guest on September 24, 2021 croenvironment (3, 4). Positive or negative selection is proposed to mune responses. Eosinophils marginated in respiratory, gut, and occur upon TCR engagement of the epithelial or hemopoietic com- urogenital subepithelium represent 95% of the total population, yet partments, respectively, based on three variables: state of thymo- their physiological role is unclear. cyte maturation, avidity of ligation, and presence of poorly defined In a study of thymic DC populations, identified by coexpression secondary signals. Thymic dendritic cells (DC)2 (2) express high of CD11c with a panel of myeloid markers, we observed a discrete levels of CD11c and MHC class II molecules and are restricted population of CD11b/CD11c double-positive cells localized primarily to the medulla (4, 5). Their role in negative selection has around the cortico-medullary region (CMR) in young mice. Isola- been established by following the fate of superantigen-reactive T tion and further characterization identified these cells as eosino- cells in bone marrow chimeric animals, in vitro by reaggregate phils. They have an activated phenotype, and their number and organ cultures (6), and through manipulations of MHC class I or II distribution vary during thymic ontogeny in a variety of mouse expression in mutant mouse models (7–9). strains. We demonstrate that eosinophils are preferentially re- The eosinophil is a terminally differentiated, end-stage granulo- cruited during class I-restricted T cell selection and that they ex- cyte that comprises Ͻ1% of circulating leukocytes (10). Conven- press presentation and costimulatory molecules. Our data indicate that the eosinophil is a numerically important member of the thymic microenvironment and imply an immunomodulatory role for *Centre National de la Recherche Scientifique, Unite´Mixte de Recherche 8603 Uni- eosinophils. versite´Paris V, and †Institut National de la Santeét de la Recherche Me´dicale, Unite´ 25, Hoˆpital Necker, Paris, France Received for publication March 8, 2000. Accepted for publication June 5, 2000. Materials and Methods The costs of publication of this article were defrayed in part by the payment of page Animals charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Male and female C57BL/6 (B6), BALB/c, DBA/2, and CBA mice were bred under specific pathogen-free conditions, which included surveillance 1 Address correspondence and reprint requests to Dr. Mark Throsby, U-BiSys, Utrecht Medical Center, Heidelberglaan 100, HP F.03.821, 3584 CX Utrecht, The for five ectoparasites and 10 endoparasites, in our own facility and used ␤Ϫ/Ϫ ␤ Netherlands. E-mail address: [email protected] between 14–21 days unless stated otherwise. Female I-A and 2- microglobulin (␤ m)Ϫ/Ϫ mice on a B6 background were also maintained in 2 Abbreviations used in this paper: DC, dendritic cell; CMR, cortico-medullary re- 2 ␤ ␤ our facility and used at 2–4 wk of age. Hemagglutinin-specific (HA) TCR gion; 2m, 2-microglobulin; B6, C57BL/6; HA, hemagglutinin; Rag, recombinase- activating gene; TNT, Tris/saline/Tween 20; AMCA, 7-amino-4-methylcoumarin-3- transgenic mice (ABII) on a BALB/c background were class II restricted acetic acid; MSS, monovalent salt solution; SA, streptavidin; FSC, forward scatter; and provided by A. Sarukhan (Institut National de la SanteétdelaRe- SSC, side scatter; MBP, major basic protein; SS, subcapsular sinus; INSERM, Institut cherche Me´dicale (INSERM), Unite´373, Institute Necker, Paris, France; Ϫ Ϫ National de la Santeét de la Recherche Me´dicale. Ref. 17). TCR H-Y mice on a Rag-1 / , Thy 1.1, B6 background were

class I restricted and provided by B. Rocha (INSERM 345, Institute g/cm3 with monovalent salt solution (MSS)-EDTA (168 mM NaCl, 42 mM Necker) (18). B. Lucas (INSERM, Unite´345, Institute Necker) provided KCl, 5 mM HEPES, and 5 mM EDTA) as previously described (5). RelBϪ/Ϫ mice on a B6 background (19). Both TCR transgenic and Thymuses from 10–20 young mice were washed and placed in RPMI RelBϪ/Ϫ mice were kept under specific pathogen-free conditions at the 1640 supplemented with 5 mM HEPES, 2% FCS, 1 mg/ml collagenase Faculty of Medicine, Hoˆpital Necker (Paris, France). Mice were injected (type III; Life Technologies), and 20 ng/ml DNase (Sigma, St. Quentin i.p., as described previously (20), with 50 nmol of peptide diluted in PBS. Fallavier, France). Tissues were first chopped with fine scissors and con- Peptide sequences were SFERFEIFPK representing aa 110–119 of HA, tinuously agitated with pipetting for 20 min at room temperature. Undi- and KCSRNRQYL, representing acids 738–746 of the Y-linked protein gested fragments were removed by unit gravity sedimentation, mechani- Smcy (21) The animal facilities and care followed the norms stipulated by cally dispersed, and recombined before addition of 0.1 M EDTA to break the European Community for the care and use of laboratory animals. up rosetting. Cells were washed through FCS-EDTA, resuspended in Ny- codenz, and centrifuged at 1700 ϫ g. Fifty to 70% of the top layer was Antibodies taken, diluted 5-fold with MSS supplemented with 5% FCS-EDTA (MSS- FCS-EDTA), and centrifuged. Cells were resuspended in MSS-FCS-EDTA The following affinity-purified mAbs were prepared in the laboratory un- on ice and incubated with the mAbs Mel 14, RA3-6B2, and KT3 for 30 less otherwise stated and titrated by FACS. For depletion, RA3-6B2 (anti- min. After washing, cells were incubated with anti-rat magnetic beads (Dy- CD45R; B220) and KT3 (anti-CD3), gifts from P. Leenen (Erasmus Uni- nal, Oslo, Norway) at a cell to bead ratio of 1:6 for 20 min at 4 C with versity, Rotterdam, The Netherlands), and Mel 14 (anti-CD62L) and YTS ␣ gentle agitation. Bound cells were removed under a magnetic field from the 169.4 (anti- CD8 ), gifts from F. Lepault (Centre National de la Recherche cell suspension, washed, and resuspended in MSS-FCS-EDTA for further Scientifique, Unite´ Mixte de Recherche 8603, Hopital Necker, Paris, selection. France) were used; for positive selection, biotinylated N418 (anti-CD11c) Positive selection of cell populations based on one phenotypic marker or M1/70 (anti-CD11b; Mac-1) was used. For cell sorting, PE-conjugated was conducted using streptavidin (SA)-conjugated MACS beads (Miltenyi HL3 (anti-CD11c), biotin-conjugated anti-CD44 (PharMingen, San Diego, b Biotec, Bergisch Gladbach, Germany). Cells resuspended in MSS-FCS- CA), and FITC-conjugated 25-7-9 (anti-I-A ) or 14-4-4S (anti-I-E), gifts EDTA were incubated with a biotinylated mAb on ice for 30 min. The cells Downloaded from from C. Boitard (INSERM, Unite´ 342, Hopital Saint Vincent De Paul, were washed, incubated with SA-magnetic beads for 30 min, washed again, Paris, France), were used. For phenotyping, anti-CD4-PerCP, CD8␣-allo- ␣ and passed through a MACS Vario column according to the manufacturer’s PE, CD8 -FITC, CD19-PE, CD11b-PE, CD25-biotin, CD49d, and CD69- instructions. Before placing cells in culture, they were washed extensively biotin CD161-PE were all purchased from PharMingen. Anti-CD80-PE and in large volumes of EDTA-free full balanced saline solution containing CD86-PE (Serotec, Oxford, U.K.), BM8 (Bachem, Voisins-le Bretonneux, 5% FCS. France), FITC-conjugated RB6-8C5 (anti-Ly-6G, Gr-1), FITC-conjugated F4/ ␥ Electronic sorting of cells was conducted after negative selection. Cells 80, and 2.4G2 (anti-Fc RII and III; gift from F. Lepault), and RM153 (anti- resuspended in MSS-FCS-EDTA were stained in three colors: anti-I-Ab or CD153/CD30L; gift from K. Okumura) were also used. Biotin-conjugated http://www.jimmunol.org/ b b I-E-FITC, anti-CD11c-PE, and anti-CD44-b followed by SA-conjugated 28-8-6 (anti-H-2K /H-2D ) was a gift from B. Rocha. ER-BMDM1 (anti- allo-PE (PharMingen). Cells were first gated based on forward (FSC) and CD13), NLDC-145 (anti-DEC-205), ER-HR3, MOMA2, and CDR1, gifts side (SSC) scatter, then on high expression of CD44. Populations were P. Leenen, were used as supernatant unless specified otherwise. For im- sorted based on their expression of MHC class II and CD11c. munohistochemistry, rabbit anti-fibronectin and rabbit anti-laminin were obtained from Novotec (Lyon, France) and rabbit anti-von Willebrand factor was purchased from Dako (Trappes, France). M1/70 (anti- CD11b/Mac-1) and FITC-conjugated N418 (anti-CD11c) were prepared in RT-PCR the laboratory. RT-PCR was conducted using conventional methodology. Briefly, cells (1 ϫ 106) in suspension were spun down and resuspended in 200 ␮lof

Triple immunofluorescent staining RNABle (EuroBio, Les Ulis, France). After separation in the presence of by guest on September 24, 2021 ␮ Thymuses were frozen in dry ice-cooled isopentane. Sections were cut at chloroform, the aqueous phase was precipitated in 2 vol of ethanol and 2 l ␮ Ϫ of pellet paint (Novagen, Madison, WI). Samples were resuspended in 3 m and air-dried. Tissues were fixed in acetone kept to 20 C for 10 ␮ min, immediately washed in two changes of PBS, then incubated with 1% 20 l of sterile RNase-free water. Reverse transcription (avian myeloblas- tosis virus, Promega, Charbonnieres, France) was performed on 5 ␮lof H2O2 for 15 min. Sections were rinsed in PBS and passed into 100 mM Tris-HCl (pH 7.6), 150 mM NaCl, and 0.05% Tween 20 (TNT) and pre- extract using random hexamer and oligo(dT) primers (Promega). Each reaction was diluted 10-fold and stored at Ϫ20 C before PCR amplification. incubated for 30 min with blocking buffer (NEN, Paris, France). Primary ␮ ␮ Abs were optimally diluted in blocking buffer (N418-FITC was diluted For PCR, 5 l of diluted cDNA was used as a template in 50- l reactions 100-fold greater than optimal) and incubated with sections for 30 min at under standard thermal cycling conditions. Reaction products were separated on 1.8% agarose gels containing 100 ng/ml ethidium bromide. Primers (5Ј to room temperature. In some staining, supernatants were applied undiluted Ј after sections had been briefly rinsed, for a further 30 min. Slides were 3 ) were: CD11b: forward, TTACTTGCGACCAGGACAGG; reverse, Ј CGTTTTCACCATCTTCTTTG (673 bp); DEC-205: forward, AGCACCGC washed in TNT, incubated with peroxidase-conjugated anti-FITC (F(ab )2; Roche, Meylan, France) for 30 min, and then washed. Sections were then CTCTTTCACCTG; reverse, TGTCCTCTTTTCCCGTAATG (600 bp); incubated for 5 min with tyramide-FITC (NEN) diluted 1/75 in amplifica- RelB: forward, CTGCGGGAGGTGGAGGTGAC; reverse, TCGTAGGGT ␣ tion buffer. After washing with agitation, sections were incubated with GGCGTTTTGAA (578 bp); IL-1 : forward, CAGTTCTGCCATTGAC swine anti-rabbit conjugated with 7-amino-4-methylcoumarin-3-acetic acid CATC; reverse, TCTCACTGAAACTCAGCCGT (218 bp); IL-2: forward, (AMCA; Vector, Burlingame, CA) and donkey anti-rat (The Jackson Lab- GACACTTGTGCTCCTTGTCA; reverse, TCAATTCTGTGGCCTGCTTG oratory, Bar Harbor, ME) conjugated with Alexas-546 (Molecular Probes, (227 bp); IL-3: forward, GACCCTC TCTGAGGAATAAG; reverse, CTC Leiden, The Netherlands) according to the manufacturer’s instructions. CAGATCGTTAAGGTGGA (232 bp); IL-4: forward, TCGGCATTTTGAA Sections were washed, mounted in PBS, and viewed and photographed on CGAGGTC; reverse, GAAAAGCCCGAAAGAGTCTC (216 bp); IL-5: for- a Leica DM epifluorescent microscope (Deerfield, IL) equipped with nar- ward, TCACCGAGCTCTGTTGACAA; reverse, CCACACTTCTCTTTTT row band filter sets optimized for Coumarin (AMCA), FITC, and Cy3 GGCG (201 bp); IL-6: forward, GTTCTCTGGGAAATCGTGGA; reverse, (Alexas-546 equivalent). TG TACTCCAGGTAGCTATGG (208 bp); IL-10: forward, AGAGCAAG The specificity of staining was routinely controlled by the omission of GCAGTGGAGCAG; reverse, GGGATGACAGTAGGGGAACC (254 bp); primary or secondary Abs or occasionally by the addition of irrelevant Abs. IL-12 p40: forward, AAGCACGGCAGCAGAAGAATAAA; reverse, All primary Abs were raised from different species, and all secondary Abs CCAACC AAGCAGAAGACAGC (478 bp); IL-13: forward, GACCCA were affinity purified and preabsorbed. Reagent dilutions were adjusted to GAGGATATTGCATG; reverse, CCAGCAAAGTCTGATGTGAG (214 balance fluorescent intensity and abrogate bleeding across filters. Images bp); IL-16: forward, AGGGACAGAACAGGGTGAGA; reverse, GTGAG were digitally acquired using a 35-mm slide scanner (Eastman Kodak, GTGGGCAGCAGAGAC (272 bp); GM-CSF: forward, TGAACCTCCTG ␣ Rochester, NY), and figures were prepared using Photoshop software GATGACATG; reverse, GTGTTTCACAGTCCGTTTCC (218 bp); TNF- : (Adobe, San Jose, CA) forward, TCTCATCAGTTCTATGGCCC; reverse, GGGAGTAGACAAGG TACAAC (212 bp); TGF-␤: forward, ACCGCAACAACGCCATCTAT; re- ϩ ␥ Isolation of CD11c thymic cells verse, GTAACGCCAGGAATTGTTGC (200 bp); IFN- : forward, GCTCT GAGACAATGAACGCT; reverse, AAAGAGATAATCTGGCTCTGC (227 To enrich the stromal elements of the thymus, a previously described den- bp); eotaxin: forward, TTCTATTCCTGCTGCTCACG; reverse, CTGGAC ␤ sity cut separation procedure was used with several modifications (5). Me- CCACTTCTTCTTCTTGG (227 bp); 2m: forward, TGACCGGCTTGTAT dia were adjusted to be iso-osmotic with mouse serum. Nycodenz (Life GCTATC; reverse, CAGTGTGAGCCAGGATATAG (222 bp); and major Technologies, Cergy-Pontoise, France) was prepared at a density of 1.069 basic protein (MBP; 387 bp) (22). The Journal of Immunology 1967

Culture conditions the medulla were also single positive (Fig. 1c, open arrows). The Cells were cultured in modified RPMI 1640 iso-osmotic with mouse serum intensity of CD11c staining was much lower on CD11b double- and supplemented with 5 mM HEPES, 10Ϫ5 M 2-ME, 1 mM sodium pyru- positive cells (Fig. 1, d and e) than on single-positive cells; note ϩ vate, 2 mM glutamine, and 100 U of streptomycin and penicillin. After also the alignment of CD11b cells at the cortical face of the extensive washing to remove EDTA, cells were dispersed in culture me- CMR. Higher power examination of the CMR (Fig. 1f) showed dium containing GM-CSF (20 ng/ml; R&D Systems, Abingdon, U.K.) and CD11bϩ/CD11cϩ cells scattered around, but distinct from, a von cultured at 106 cells/ml in 24- or 96-well plates. Willebrand factor-positive vessel. CD11b single-positive cells Flow cytometry were also observed, but were directly adjacent to the venules in ϩ ϩ For three-color analysis, 106–105 cells were stained with FITC-, PE-, and perivascular spaces. The distribution of CD11b /CD11c cells biotin-conjugated mAb in MSS-FCS-EDTA, washed, and incubated with strongly suggests they are part of the tissue parenchyma. Double SA-allophycocyanin. Samples were washed in MSS-EDTA and 2% FCS labeling with anti-CD11b and anti-fibronectin shows that most and were analyzed immediately without fixation on a FACScalibur (Becton CD11b-positive cells in the CMR and medulla are proximal to, but Dickinson, Mountain View, CA). Data were analyzed with WinMDI software (http://facs.scripps.edu/). distinct from, fibronectin-positive extracellular matrix structures (Fig. 1g); a similar pattern of staining was observed for laminin Immunohistochemisty (not shown). In contrast, the SS (inset) and septa (not shown) con-

Tissue sections were fixed and preincubated with H2O2 and blocking buffer tained CD11b-positive cells embedded in fibronectin-positive con- as described for immunofluorescent histochemistry. Primary Ab were ap- nective tissue. At high power (Fig. 1h), CD11bϩ/CD11cϩ cells in plied either as supernatant or optimally diluted in blocking buffer and in- the CMR were heterogeneous in terms of morphology, but ap- cubated at room temperature for 30 min. Slides were washed in TNT and peared multinucleated in many cases (inset). Staining with myeloid Downloaded from incubated with peroxidase-conjugated anti-rat IgG or anti-hamster IgG (The Jackson Laboratory) for an additional 30 min. Labeling was visual- markers, including the rat anti-mouse macrophage markers ized with diaminobenzene in the presence of nickel chloride. ER-HR3 (Fig. 1i), MOMA2 (Fig. 1j), F4/80, HR-MP23, BM8, and MOMA1 (not shown), showed no comparable pattern of labeling TUNEL to that of CD11b. To investigate further the phenotypic character- Cryostat sections of 4 ␮m were fixed in 4% paraformaldehyde for 10 min istics of the CD11bϩ/CD11cϩ population, we isolated CD11cϩ at room temperature, washed in PBS, and permeabilized in ethanol/acetic

cells from the thymus. http://www.jimmunol.org/ acid (2/1) for 5 min. Slides were washed in PBS and preincubated in reaction buffer (100 mM sodium cacodylate (pH 6), 1 mM CoCl2, 0.1 mM DTT, and 0.005% BSA) for 10 min at 37 C. Sections were blotted, and ϩ reaction buffer containing 20 U of TdT, 20 ␮M dNTP mix (Promega), and Selection and phenotyping of thymic CD11c cell populations 2 ␮M digoxigenin-11-dUTP (Roche) was added for 60 min at 37°C. Slides ex vivo were then passed into 2ϫ SSC for 30 min at 37°C, washed in PBS, and ϩ Ј To obtain the CD11c thymic population, collagenase-digested incubated with alkaline phosphatase-conjugated anti-digoxigenin (F(ab )2; Roche) for 30 min. After further washing, slides were placed into substrate tissue was fractionated by density gradient centrifugation (5), and buffer (100 mM Tris (pH 9.5), 150 mM NaCl, and 50 mM MgCl2) for 10 the low buoyant density fraction (6–8% of total) was depleted of min. Sections were then incubated in substrate buffer containing 4.5 ␮g/ml Mel 14ϩ (anti-CD62L), CD3ϩ, and B220ϩ (anti-CD45R) cells by

␮ by guest on September 24, 2021 4-nitroblue tetrazolium chloride and 3.5 g/ml 5-bromo-4-chloro-3-in- magnetic beads (ϳ60% CD11cϩ). They could be further enriched dolyl-phosphate and monitored for intensity of staining. Sections were washed in PBS containing 10 mM EDTA and mounted. with positively selecting MACS beads or stained in three colors and electronically sorted for further analysis. Measurement of NADPH oxidase and peroxidase activities All CD11cϩ cells express high levels of CD44 (Fig. 2) and The methodology and reagents were described in detail previously (23). heat-stable Ag (not shown), but could be split into two populations Briefly, eosinophils were enriched from monodispersed thymocytes as de- based on the level of class II expression (Fig. 2b). The population scribed above. The Abs NLDC-145 (anti-DEC-205) and YTS 169.4 (anti- ␣ characterized by intermediate expression of CD11c and low to CD8 ) were added to the depletion mix, in addition to Mel 14, KT3, and ␣Ϫ RA3-6B2, to remove DC. Alternatively, eosinophils and DC were sorted to negative class II, expressed CD11b and was CD8 . In contrast, high purity as described above. After purification steps, all cells were washed the class II population had higher expression of CD11c and was extensively in HBSS without calcium or magnesium, brought to room tem- CD8␣ϩ CD11bϪ, consistent with the previously reported pheno- perature, and supplemented with calcium. Basal and PMA-stimulated (27 type of thymic DC (5). The low buoyant density depleted thymic nM) NADPH oxidase (O oxidoreductase) activities were measured over 2 cells were electronically sorted as described in Materials and 60 min as the luminescence product of dimethylbiacridinium (lucigenin) Ͼ reductive deoxygenation. Basal and PMA-stimulated (2.7 ␮M) peroxidase Methods (Fig. 2c); the sorted population was routinely 98% pure ϩ (H2O2 oxidoreductase) activities were measured as the luminescence prod- on reanalysis. Microscopic examination of sorted CD11b cells uct of luminol-deoxygenation. after May-Gru¨nwald Giemsa staining revealed an annular or con- voluted nucleus with a small irregular cytoplasm. The presence of Results eosinophilic granules in the cytoplasm identified the CD11bϩ cells A population of CD11b/CD11c double-positive cells localizes to as eosinophils (Fig. 3b). Cultured eosinophils were dependent on the CMR GM-CSF in culture. They put out a heterogeneous array of pro- Triple-labeling immunohistochemical studies were conducted in jections (that could sometimes be observed ex vivo; Fig. 3c), rang- 14-day-old C57BL/6 mice using a panel of myeloid markers, anti- ing from thick to thin very long tendril-like strands, and were CD11c (N418), and either the endothelial marker von Willebrand actively motile (Fig. 3d). factor or fibronectin to stain the extracellular matrix. Fig. 1, a and Phenotypic analysis was conducted on CD11c-selected popula- b, illustrates the demarcation between cortical epithelial cells and tions gated for eosinophils or DC based on scatter characteristics medullary CD11cϩ DC. At this age, we observed a striking pattern and expression of class II and CD11c (Fig. 4a). Eosinophils ex- of staining for anti-CD11b around the CMR and subcapsular sinus press myeloid-related markers, including F4/80 (24) and the more (SS). In triple-labeling experiments most CD11bϩ cells colocal- restricted macrophage marker BM8, but not CD13. Low surface ized with CD11cϩ cells, predominately around the CMR but also expression of Fc␥RII, Fc␥RIII, CD25, and CD69 was observed in small numbers inside the medulla and cortex (Fig. 1c, white consistent with previously reports of activated eosinophils (22, arrows). However, CD11bϩ cells rarely stained for CD11c in the 25). In contrast, eosinophils did not express the thymic DC mark- inter- and intralobular septa and SS. A small number of cells inside ers DEC-205 (Fig. 4a), CD8␣ (Fig. 2b) or BP-1 (5) (not shown); 1968 EOSINOPHILS IN THE MURINE THYMIC MICROENVIRONMENT Downloaded from http://www.jimmunol.org/ by guest on September 24, 2021

FIGURE 1. CD11b/CD11c double-positive cells are preferentially localized around CMR in the young murine thymus. Acetone-fixed cryostat sections were stained by immunofluorescence for N418 (CD11c; a–d, f, and h–j), M1/70 (CD11b; c and e–h), NLDC-145 (DEC-205; a), CDR1 (cortical epithelium; b), HR3 (macrophage; i), MOMA2 (macrophage/DC; j), anti-von Willebrand factor (endothelium; a–c, f, and h–j), anti-fibronectin (extracellular matrix; g). a, Seen at low power, staining with NLDC-145 (Alexas-546, red) strongly labels the cortical epithelium; staining of CD11c (FITC, green)-positive DC is not obvious (they are positive by FACS analysis). b, The reticular staining of the cortical epithelial marker CDR1 (Alexas-546, red) contrasts with the more confluent staining of CD11c in the CMR; note the presence of venules stained in blue. c, CD11b (Alexas-546, red) single-positive cells are found in the subcapsular sinus and septa, but rarely in the CMR (unfilled arrows). Discrete CD11bϩ/CD11cϩ staining (yellow) is observed around the CMR and The Journal of Immunology 1969 the granulocyte marker Ly-6G (Gr-1); or the myeloid differentia- CD11cϩ eosinophils are inside the tissue parenchyma, and their tion/activation marker Ly-6C (not shown). Neutrophils and mono- NAPDH oxidase activity is maximal under nonstimulated cytes generally express the latter at high levels. conditions. Eosinophils can act as APC and express costimulatory molecules under activating conditions (11, 14, 15). Thymic eosinophils Two distinct phases of eosinophil recruitment occur during express class II molecules (Fig. 2b) and intermediate levels of class postnatal developmental of the thymus I molecules (Fig. 4a). Low surface expression of the costimulatory During the neonatal period, the thymus undergoes expansion as molecules, CD86 (B7.2) and CD30L (CD153), a membrane-bound cells are generated to furnish the adult peripheral T cell pool; new member of the TNF family, but not CD80 (B7.1), was detected production is diminished with age, as reflected in shrinking of the (Fig. 4a). Thus, thymic eosinophils are likely to be able to cortex (4). Eosinophil numbers increase sharply during the neona- present Ag. tal period to reach a maximum at 2 wk of age (Fig. 6a). At this time point they are roughly equal in absolute number to DC iso- Pattern of cytokine expression in sorted thymic eosinophils lated in tandem. While the number of DC continues to increase Previous studies of eosinophil cytokine expression suggest that with tissue size, the absolute number of eosinophils decreases, re- different combinations of cytokines may be linked to activation or flected by a sharp drop in eosinophil numbers at 28 days of age disease states (26). We analyzed total RNA extracted from popu- (Fig. 6b). Absolute numbers of eosinophils begin to rise again at lations of eosinophils and DC, sorted to 98% purity (Fig. 2c), for 16 wk, corresponding to the commencement of thymic involution. cytokine transcripts by RT-PCR (Fig. 4b). A strong signal was The localization of eosinophils also changes during ontogeny. In ϩ ϩ present for the eosinophil-restricted granule component MBP in neonatal animals, CD11b /CD11c cells are concentrated at the Downloaded from the CD11bϩ/CD11cϩ double-positive population. Eosinophils ex- CMR (Fig. 1); at latter time points they become more prominent in pressed TGF-␤ and IL-16 mRNA consistent with their wide dis- the medulla (not shown). No differences were noted in cell number tribution among leukocytes. Detectable mRNA levels of the proin- or phenotype between sexes or among C57BL/6, BALB/c, DBA/2, flammatory cytokines, IL-1␣, IL-6, and TNF-␣ were present as and CBA strains of mice. In RelBϪ/Ϫ mice, who lack thymic med- previously reported in activated eosinophils (14, 15). Expression ullary epithelial cells and thus fail to form a medulla (19), few ϩ of eotaxin, IL-2, IL-3, IL-10, IL-12, IFN-␥, GM-CSF, and IL-5 CD11b cells were detected in the thymic parenchyma compared http://www.jimmunol.org/ was undetectable. GM-CSF and IL-5 has been reported to act as with age-matched RelBϩ/Ϫ littermates (1.2 Ϯ 0.2 vs 51.7 Ϯ 2.7 autocrine survival and recruitment factors for activated eosinophils cells/mm2, mean Ϯ SEM; n ϭ 3; p Ͻ 0.001, by independent t test), in inflammatory foci (12). However, thymic eosinophils expressed although they were observed in the SS (not shown). mRNA for the closely related Th2 cytokines IL-4 and IL-13. Both are linked to eosinophil involvement in certain pathologies (27, 28) Acute negative selection in class I-restricted TCR transgenic and are reported to aid recruitment, activation, and survival (29, 30). mice is associated with eosinophil recruitment Transcriptional expression of CD11b was readily detectable in Given the localization of eosinophils around the CMR, in contact both eosinophils and DC (Fig. 4b), although low surface expres- with immature double-positive thymocytes susceptible to negative sion is present on a minority of freshly isolated DC (Fig. 2b) (5). selection, and their augmented recruitment during the neonatal pe- by guest on September 24, 2021 DC, but not eosinophils, express the scavenger receptor DEC-205, riod, we examined their behavior in a model of acute negative consistent with FACS profiles, and eosinophils did not express selection. Injection of the cognate peptide into TCR transgenic relB, a nuclear protein of restricted expression that regulates mice induces prompt apoptosis in thymocytes (20, 32–34). While NF-␬B activity in DC and medullary epithelial cells (19). at later time points thymic atrophy can be linked to peripheral activation (32, 33), initially cell death is due to peptide presenta- Basal superoxide anion production by thymic eosinophils tion within the thymus (33, 34). Recently it was observed that We investigated whether functional activation accompanied the peptide injection into a class I-restricted TCR transgenic model presence of activation markers. Fig. 5a shows a representative ex- resulted in an infiltration of CD11bϩ granulocytes into the thymus periment in which Ͼ60% of depleted cells are eosinophils. The within 1 h. This was before apoptosis became apparent, and no level of superoxide anion released from total thymocytes, using other organs were affected (20). We repeated the experiments on lucigenin-amplified chemiluminescence, was marginally above class I- and class II-restricting backgrounds. baseline. In the low buoyant density fraction, there was a log fold The proportion of total thymocytes stained for annexin V in increase and 2 log fold increases in the depleted population. The class I-restricted female H-Y TCR transgenics increased 6 h after time-dependent profile of luminescence emission is indicative of cognate peptide injection, particularly among the CD4low, CD8low NADPH oxidase activation. The peak level of superoxide anion double-positive population associated with apoptosis (Fig. 7a) release in 98% pure FACS-sorted populations (Fig. 2c) was not (35). This was reflected in a significantly higher number of significantly different after PMA stimulation; however, peroxidase TUNEL-positive apoptotic bodies in cryostat sections from pep- activity was significantly increased (Fig. 5b). It has been reported tide-treated mice (Table I). We observed a progressive increase in that eosinophil extravasion can result in heightened superoxide the number of CD11bϩ cells in the thymus after peptide injection. anion release (31), consistent with our observation that CD11bϩ/ Isolated cells had the same phenotype as thymic eosinophils from

in the medulla (solid arrows), but rarely in the cortex. d and e, Staining for CD11c (d) and CD11b (e) in the same field. Staining for CD11c (arrows) on CD11bϩ cells is faint compared with that on DC, but is still visible at lower Ab titration. f, Higher power view of the CMR. CD11bϩ/CD11cϩ cells are distributed along the cortical face near a large von Willebrand factor-positive venule. Note the small number of CD11b single-positive cells directly adjacent to the venule. g, Seen without CD11c staining, CD11b-positive cells are concentrated at the CMR and sparsely distributed inside the medulla; staining is distinct from the fibronectin-positive extracellular matrix (AMCA, blue) in contrast to positive cells in the subcapsular sinus (inset), which are distributed along fibronectin-positive connective tissue. h, At high magnification CD11bϩ/CD11cϩ cells are heterogeneous in terms of shape and size and are ap- parently multinucleated; inset, occasionally dendritic projections are stained. i, HR3 (Alexas-546, red) strongly labels macrophages in the cortex. j, MOMA2 (Alexas-546, red) labels macrophages in the cortex and medulla. Some CD11cϩ cells are also stained. Scale bars ϭ 100 ␮m. 1970 EOSINOPHILS IN THE MURINE THYMIC MICROENVIRONMENT Downloaded from

FIGURE 3. Conventional May-Gru¨nwald-Giemsa staining of cytospins from progressively enriched populations of CD11cϩ thymic cells. a, Total http://www.jimmunol.org/ thymocytes from C57BL/6 mice; b, low buoyant density fraction from male H-Y TCR transgenic mice; c, FACS-sorted eosinophils; inset, phase contrast photograph of sorted eosinophil after 18 h culture in GM-CSF; d, sorted eosinophils after 7-day culture in the presence of GM-CSF. Orig- inal magniﬁcation, ϫ1250 (a and c) and ϫ675 (b and d).

roughly equivalent in both negatively selecting males, which de- lete all TCRϩ thymocytes, and positively selecting female mice. by guest on September 24, 2021 However, the relative percentage of eosinophils to total cellularity in males was 2-fold higher (Fig. 7b). In serial thymic sections from female H-Y TCR transgenics stained for CD11b and DNA fragmentation, a measure of apoptosis, there was a scattering of apoptotic bodies in and around medullary regions (Fig. 8a). Staining for CD11b was also concentrated in and around the medulla (Fig. 8d). Three hours after injection, vessels containing clusters of CD11bϩ cells appeared (not shown). FIGURE 2. Magnetic bead isolation and FACS sorting of CD11b/ By 6 h apoptotic bodies were distributed throughout the tissue, CD11c double-positive eosinophils from collagenase-digested thymic tis- with the exception of cortical regions directly underneath the cap- sues. Cells were prepared as described in Materials and Methods. a, Note Ϫ Ϫ sule, the site of immature TCR thymocytes (Fig. 8b). The distri- enrichment of CD44high, CD62L cells (upper panel) and CD11cint, I-Alow/Ϫ cells (lower panel) after density gradient separation in the low bution of CD11b-positive cells closely followed that of the buoyant density population (5% of total cells) and further enrichment after TUNEL staining (Fig. 8e). In male mice, in which negative selec- magnetic bead separation with Abs against CD62L, CD45R, and CD3. tion is constitutive, the level of apoptosis is not as significant as ϩ b, The depleted population was further purified by positive selection for that after peptide administration (Fig. 8c), but CD11b eosinophils CD11c (93% purity). Three populations are apparent: 1) CD11cint, are distributed throughout the tissue (Fig. 8f). Given the age-re- Ϫ Ϫ Ϫ ϩ I-Ablow/ , CD11bhigh, CD8␣ ; 2) CD11chigh, I-Abint, CD11b , CD8␣ ; lated differences in eosinophil recruitment we conducted experi- and 3) CD11cint/high, I-Abhigh, CD11bϪ, and CD8␣ϩ. c, Electronic cell ϩ ϩ ments at 14, 21, and 35 days of age. Peptide administration in- sorting of CD11b /CD11c cells from depleted, collagenase-digested thy- duced similar increases in eosinophil numbers at all ages (not mic tissue (as described in Materials and Methods). Triple-labeled cells shown). Thus, eosinophil recruitment is constitutively higher in (upper panel) were gated for high SSC (R1), high expression of CD44 (R2) female H-Y TCR mice and can be further increased by peptide intermediate staining of CD11c, and low to negative staining of class II (R3) and were sorted with a FACSvantage cell sorter. Upon reanalysis presentation. Although cellularity is considerably diminished in (lower panel) Ͼ98% of cells fell within a gate 10% larger than the sort gate male mice, in which cognate Ag is constitutively available, eosin- (R3); note the high SSC and low FSC of the sorted population. ophil numbers are at levels similar to those in females. In contrast, injection of the cognate peptide into class II-restricted HA TCR mice had no effect on the recruitment of eosin- normal mice, and cytospins of enriched populations confirmed ophils despite an increased number of apoptotic bodies, as assessed them to be the only granulocyte population. Six hours after injec- by TUNEL staining (Table I). The absolute number of CD11bϩ tion the absolute number of CD11bϩ/CD11cϩ cells had increased cells was not different from that in age-matched control strains 75% (Fig. 7b). In terms of absolute numbers, eosinophils are either before or after injection at 28 days (Fig. 7c), although the The Journal of Immunology 1971 Downloaded from http://www.jimmunol.org/

FIGURE 5. Basal and stimulated extracellular production of superoxide by guest on September 24, 2021 anion in cell populations progressively enriched for eosinophils. a, Lucige- nin-amplified chemiluminescence (CL) production was measured each min over1hinasuspension of freshly isolated total thymocytes taken from collagenase-digested tissue (dotted lines), hypodense thymic cells after density gradient centrifugation (dashed lines), and eosinophil-enriched populations after magnetic bead depletion, including NLDC-145, CD4, and CD8 to remove DC (solid lines). b, Histograms representing the peak mean counts of lucigenin-amplified CL representing NAPDH activity and luminol-amplified CL measuring peroxidase activity from basal (Ⅺ) and PMA- stimulated (f) eosinophils.

number of CD11bϩ cells measured on tissue sections was significantly lower than that in control strains at 14 days (Table I). MHC II-null mutant mice and class I-null mice contained similar numbers of eosinophils as age-matched controls (Fig. 7c). Therefore, the signals drawing thymic eosinophils may be a result of class FIGURE 4. a, Phenotypic characterization of purified thymic CD11cϩ ϩ I-restricted selection and/or may be related to the high number of eosinophils and DC. N418 cells selected by magnetic bead depletion were mature CD8 single-positive cells in the thymus of H-Y TCR mice. stained in three colors for FACS analysis. Each histogram represents 104– 105 cells acquired on a FACSCalibur and gated for characteristic FSC and SSC and differences in class II expression: eosinophils; FSClow, SSChigh Discussion I-Alow and DC; FSChigh I-Aint/high. b, RT-PCR analysis of cytokine- and Eosinophils were identified in this study by the dual expression of ϩ ϩ cell-specific protein expression from extracts of sorted thymic cell popu- CD11b and CD11c. In young mice, CD11b /CD11c cells were lations. Total RNA extracts from sorted populations (1 ϫ 105 cells) of anatomically restricted and had a granular appearance under high eosinophils (top panel) or eosinophils (Eo) and DC (lower panel) were power observation. The eosinophil specific traits, expression of ␤ reverse transcribed, cDNA diluted 1/5 (or 1/200 for 2m), and amplified MBP, and eosinophilic cytoplasmic granules after Giemsa staining with 35 cycles of PCR using primers specific for the coding sequences of characterized the sorted population. the molecules shown. Extracts from bone marrow and mitogen-activated Thymic eosinophils display some features of activated eosino- splenocytes were used as positive controls. A representative experiment of phils. There is a loss of Gr-1 and CD62L expression, which are three is given. present on bone marrow eosinophils. They express several phenotypic markers associated with activation, including CD25, CD69, 1972 EOSINOPHILS IN THE MURINE THYMIC MICROENVIRONMENT Downloaded from http://www.jimmunol.org/

FIGURE 6. Number of CD11bϩ eosinophils and CD8␣ϩ DC present during thymic ontogeny. Low buoyant density thymic cells representing Ͼ8% of the total number of thymocytes were isolated as described in Materials and Methods and stained in three colors with the Abs anti-I-Ab- by guest on September 24, 2021 FITC or anti-I-E-FITC, anti-CD11c-PE, and anti-CD11b-biotin or anti- CD8␣-biotin; streptavidin-APC was used as a second layer. CD11bϩ cells inside gates R1 and R3 (shown in Fig. 3) were considered eosinophils, CD8␣ϩ cells gated for high FSC, high expression of CD11c, and class II (see Fig. 2) were counted as DC. a, Mean number of eosinophils (F) and DC (E) isolated per thymus. b, Eosinophils (F) and DC (E) expressed as the percentage of total thymocytes. Duplicate staining was conducted and quantiﬁed from in three to eight animals; error bars represent the SEM.

FIGURE 7. FACS analyses of thymic cell populations and quantifica- and low levels of class II. The cells are hypodense, a characteristic tion of eosinophil numbers after i.p. administration of 50 nmol of cognate of some circulating eosinophils in hypereosinophilic patients and peptide into TCR transgenic mice. a, Upper panel, Total thymocytes in allergic inflammatory foci (10, 36), and their level of superoxide stained in three colors: annexin V-FITC, CD4-PerPC, and CD8-APC. Cells production was maximal without further stimulation. Although lit- (104) were acquired on a FACSCalibur. Contour plots show that 6 h after tle information is available on the phenotypic and functional char- peptide injection in H-Y TCR transgenic mice, there is a clear increase in low low acteristics of tissue-marginated eosinophils, it is proposed that par- the CD4 /CD8 population (gated), of which 60% are positive for sur- tial activation or priming may be a normal facet of eosinophil entry face annexin V (histogram inset); these cells account for about 7% of the total thymocytes ϫ 3.5 control animals (6 h of PBS). b and c, Low buoyant into tissue (10). Movement across endothelial barriers is enough to density thymic cells representing Ͼ8% of the total number of thymocytes activate circulating human eosinophils, as measured by increased were isolated as described in Materials and Methods and stained in three expression of CD11b and superoxide anion production (31), both colors with anti-I-Ab-FITC or I-E-FITC, anti-CD11c-PE, and anti-CD11b- features of thymic eosinophils. biotin; streptavidin-APC was used as a second layer. CD11bϩ cells inside RT-PCR of sorted eosinophil RNA extracts revealed a distinct gates R1 and R3 (shown in Fig. 3) were considered eosinophils. Histo- pattern of cytokine mRNA. Differential patterns of eosinophil cy- grams show the number of eosinophils isolated per thymus on the left tokine production have been reported in different disease states. ordinate and the percentage of total thymocytes on the right ordinate for For example, IL-5 is expressed by intestinal tract eosinophils in class I-restricted H-Y TCR transgenic mice (21 days old) and age-matched ␤ Ϫ/Ϫ celiac, but not Crohn’s, disease (13). Eosinophils taken from bron- controls (b) and class II-restricted HA TCR transgenic mice, 2m , ␤Ϫ/Ϫ chial lavage or nasal polyps of atopic individuals express IL-5 and I-A (28 days old) and the age-matched control strain (c). Data were collected from at least two separate experiments, and each point represents GM-CSF, which are proposed to act in an autocrine fashion, re- p Ͻ ,ء .the mean of three or more animals; error bars represent the SEM spectively, to enhance recruitment to (37) and survival (12) at sites 0.01 compared with vehicle-treated mice (n ϭ 3), result of independent of inflammation. Eosinophils are capable of producing multiple t test. subsets of cytokines in a directed fashion. Binding of IgA com- The Journal of Immunology 1973

Table I. Number of CD11bϩ cells and apoptotic bodies in thymic sections from peptide-treated TCR transgenic micea

Age Time Before No. of Fields CD11bϩ TUNELϩ Strain (days) Sex Treatment Sacriﬁce (h) per Section cells/mm2 cells/mm2

C57BL/6J 14 F – – 102 Ϯ 9.8 122 Ϯ 4.6 ND H-Y TCR 14 F Saline i.p. 6 95 Ϯ 24 201 Ϯ 27 592 Ϯ 51 H-Y TCR 14 F H-Y peptide i.p. 3 103 Ϯ 25 328 Ϯ 36* ND H-Y TCR 14 F H-Y peptide i.p. 6 78 Ϯ 35 573 Ϯ 64* 2177 Ϯ 210* H-Y TCR 14 M – – 34 Ϯ 1.4 580 Ϯ 60 631 Ϯ 32 HA TCR 14 F Saline i.p. 6 98 Ϯ 16 80 Ϯ 33 80 Ϯ 12 HA TCR 14 F HA peptide i.p. 6 101 Ϯ 21 91 Ϯ 18 211 Ϯ 29*

a Acetone-ﬁxed cryostat sections cut from the thymus of C57BL/6 mice were immunostained with anti-CD11b (M1/70), followed by anti-rat peroxidase and diaminobenzidine. TUNEL staining was carried out as described in Materials and Methods. Positively stained cells outside subcapsular sinuses or interlobular septa were counted by reticle on duplicate serial sections at three noncontiguous levels in at least two different mice. The number of reticle ﬁelds counted, corresponding to 0.11 mm2 of section, that could be placed entirely within a section is expressed as mean Ϯ SD. The number of CD11bϩ or TUNELϩ cells staining per mm2 of section is expressed as mean Ϯ SEM. *, p Ͻ 0.01 compared with vehicle-treated mice, n ϭ 3, result of independent Students t test.

plexes in vitro induces IL-10 secretion in eosinophils taken from of circulating (39) and tissue-marginated eosinophils, including Downloaded from hypereosinophilic patients, while the ligation of CD28 induces thymic eosinophils (40), suggesting that the baseline level of eo- IL-2 and IFN-␥ secretion (11). sinophils in the thymus is maintained through an eotaxin-driven Thymic eosinophils expressed mRNA for IL-4 and IL-13, which pathway of extravasion. are closely related cytokines that share receptors and have over- Comparison of CD11cϩ cells isolated at different time points lapping functions. Critical in generating eosinophil responses to suggests that eosinophil recruitment is particularly augmented be- parasitic infection (27), IL-4 and IL-13 have recently been asso- tween 7–14 days and in older mice (Ͼ100 days old). Recruitment http://www.jimmunol.org/ ciated with activation and survival of human peripheral blood eo- seems to diminish between these periods, since the absolute num- sinophils (29) and recruitment of murine eosinophils during aller- ber of thymic eosinophils drops 50% between 14 and 28 days of gen-induced airway inﬂammation (30), both in synergy with age. In comparison, DC numbers are closely related to total thymic TNF-␣. Expression of IL-4, IL-13, and TNF-␣, therefore, repre- cellularity, consistent with experiments showing that both T cells sents a cytokine pattern associated with extravasion and activation. and DC arise from the same progenitor in the thymus (41). These It seems likely that eosinophils are recruited into the thymus as data imply that eosinophils have a relatively short half-life in the in normal subepithelial mucosa. Eotaxin, an eosinophil-speciﬁc thymus; it has been calculated that marginated tissue eosinophils CCR3 ligand, is constitutively expressed in thymus and lymph survive for at least 3 wk after extravasion (10). On tissue sections nodes (38). Eotaxin null mice have a reduction in both the number there is also a distinct difference in anatomical distribution; by guest on September 24, 2021

FIGURE 8. Distribution of CD11bϩ cells in the thymus after cognate peptide injection in female H-Y TCR transgenic mice and in untreated male mice. Immunohistochemical staining was conducted on serial sections for TUNEL labeling of fragmented DNA, a marker of apoptosis (a–c), and anti-CD11b (M1/70; d–f). In tissue from PBS-injected H-Y TCR female mice (a and d) scattered apoptosis was also observed, particularly around medullary regions where CD11bϩ cells are concentrated. b and e, Six hours after peptide injection apoptotic bodies were present throughout the tissue; staining for CD11bϩ was distributed throughout ﬁelds of apoptotic bodies. c and f, In male mice apoptotic bodies were scattered throughout the tissue, although not at the same density as in the acute negative selection model, CD11b staining was observed concentrated in vessels and spread throughout the tissue. c, cortex; sc, subcapsular sinus; m, medull. Scale bars ϭ 100 ␮m. 1974 EOSINOPHILS IN THE MURINE THYMIC MICROENVIRONMENT

2-wk-old animals have CD11bϩ cells concentrated in the CMR, ysis in particular pathologies, e.g., hypereosinophilia, or in hyper- while at 16 wk eosinophils are distributed evenly throughout the sensitivity reactions (10). These situations may not accurately medulla and into the CMR. reflect the physiological role of eosinophils, especially tissue-mar- Without knowing the source of the chemotactic signal or its ginated eosinophils, under nonpathological conditions. In this stimulus, it is difficult to determine the significance of the differ- study we show that murine thymic eosinophils express Ag-pre- ential spatial and temporal distribution of thymic eosinophils. Sev- senting molecules and have several physical characteristics of eral lines of evidence suggest that under physiological conditions, APCs that enhance cellular contact, including nonadherence, mo- negative selection of developing double-positive thymocytes oc- tility, and dendritic projections. Others have shown that both hu- curs in the CMR (2, 4). In addition, between 7 and 14 days of age man and murine eosinophils are capable of acting as APCs. They there is an exponential increase in the cellularity of the thymus express costimulatory molecules previously shown to be involved caused primarily by the clonal expansion of double-positive thy- in clonal deletion, such as CD30 ligand (CD153) and CD86 mocytes (1). Accordingly, recruitment of eosinophils is associated (44, 45). Uniquely, eosinophils produce considerable levels of free with increased demand for repertoire selection and is localized to radicals that diffuse freely through membranes in the immediate the region in which it is occurring. vicinity of their release. Developing thymocytes may have an Thus, an important observation in this study was the dramatic increased sensitivity to these molecules (e.g., through down-regu- elevation of eosinophil numbers in a model of acute negative se- lation of Cu/Zn superoxide dismutase) that could induce apoptosis lection. This finding confirms and extends a previous report from in concert with other signals given simultaneously or sequentially. a different MHC class I-restricted TCR transgenic line in which Eosinophils have been identified in the thymus, based on their ϩ cognate peptide injection induced a prompt influx of CD11b distinctive granular staining (40, 46, 47) or peroxidase activity Downloaded from granulocytes (20). In both studies, CD11bϩ cells were distributed (48). In this study we describe a method by which tissue-bound throughout the apoptotic fields. The granulocyte recruitment is un- eosinophils can be isolated from normal tissue. We demonstrate likely to be an artifact of systemic activation, since infiltration was that the eosinophil is a regulated component of the murine thymus already significant at 3 h, was only apparent in the thymus, and that is recruited in the absence of overt inflammatory stimulus was selective for eosinophils. In addition, eosinophils were dis- similar to other tissue-marginated eosinophils. In mucosal tissue,

tributed throughout the tissue of the negatively selecting male H-Y eosinophils assume a sentinel position in the subepithelium among http://www.jimmunol.org/ TCR transgenic mouse in which the cognate Ag is constitutively macrophages, DC, and lymphocytes (40). Their precise role under present. When these experiments were repeated in class II-re- these circumstances is not clear; however, as with thymic eosino- stricted HA TCR mice, systemic administration of cognate peptide phils, it is likely to be different from that of eosinophils recruited failed to recruit eosinophils, although significant apoptosis was during allergic inflammation. Further insight into this cell popula- observed. tion may be useful in clarifying not only the role of eosinophils in Taken together, eosinophil recruitment above baseline is asso- thymic function, but also the physiological role of tissue-margin- ciated with high affinity, class I-restricted selection/deletion and is ated eosinophils in general, leading to a better understanding of not likely to be a nonspecific reaction to cell death. However, their participation in a variety of physiological and pathological significant differences exist between TCR transgenic models, in responses, in particular atopic diseases. by guest on September 24, 2021 particular the affinity of the TCR for the peptide/MHC complex, that may affect the mechanism of thymic deletion (42). Although Acknowledgments striking differences in eosinophil numbers are observed between We acknowledge A. Esling for technical assistance, I. Cisse for animal H-Y and HA TCR transgenic models we cannot exclude the pos- husbandry, C. Garcia (Institut National de la Sante´ et de la Recherche sibility that these might be intrinsic to the model and not a result Me´dicale, Unite´373) for assistance with cell sorting, and A. T. Nguyen of the restriction element. Further analysis of TCR transgenic mice (Institut National de la Santeét de la Recherche Me´dicale, Unite´507) for that encompasses a wide spectrum of avidity will determine chemiluminescence measurement. P. J. Leenen, F. Lepault, and C. Boitard whether eosinophil recruitment is indeed associated with MHC are thanked for the kind gift of antiserum; M. Dy for cytokine PCR prim- class I-restricted selection or with a particular mechanism of thy- ers; B. Rocha, B. Lucas, and A. Sarukhan for the use of their TCR trans- mocyte deletion. genic lines; and P. Throsby for advice with figures. We are gratefully to B. Rocha, F. Geissmann, R. Monteiro, B. Descamps-Latscha, A. Sarukhan, Interestingly, the number of thymic eosinophils in female Rag- and M. Dy for advice on the manuscript. 1Ϫ/Ϫ H-Y mice was greater than that age-matched controls. These mice have an increased thymic medullary area and an exaggerated proportion of mature CD8ϩ cells compared with normal mice. References 1. Goldrath, A. W., and M. J. Bevan. 1999. Selecting and maintaining a diverse Therefore, signals released from mature thymocytes or other med- T-cell repertoire. Nature 402:255. ullary components could elevate eosinophil numbers under steady 2. Nossal, G. J. 1994. Negative selection of lymphocytes. Cell 76:229. state conditions. Consistent with this hypothesis, eosinophils were 3. van Ewijk, W. 1991. T-cell differentiation is influenced by thymic microenvi- Ϫ/Ϫ ronments. Annu. Rev. Immunol. 9:591. not present in RelB mice, which do not form a thymic medulla 4. Anderson, G., N. C. Moore, J. J. Owen, and E. J. Jenkinson. 1996. Cellular but are characterized by granulopoiesis (19). Mature CD8 single- interactions in thymocyte development. Annu. Rev. Immunol. 14:73. positive cells only begin to populate the medulla after 7 days (43), 5. Vremec, D., M. Zorbas, R. Scollay, D. Saunders, C. Ardavin, L. Wu, and K. Shortman. 1992. The surface phenotype of dendritic cells purified from mouse consistent with the time point at which a large influx of eosinophils thymus and spleen: investigation of the CD8 expression by a subpopulation of is observed. dendritic cells. J. Exp. Med. 176:47. Recent evidence has cast the traditional nonredundant physio- 6. Jenkinson, E. J., G. Anderson, and J. J. Owen. 1992. Studies on T cell maturation on defined thymic stromal cell populations in vitro. J. Exp. Med. 176:845. logical role of eosinophils, protection against helminthic infection, 7. Brocker, T., M. Riedinger, and K. Karjalainen. 1997. Targeted expression of into doubt (10). Moreover, eosinophils express a diverse array of major histocompatibility complex (MHC) class II molecules demonstrates that dendritic cells can induce negative but not positive selection of thymocytes in cytokines and cell surface ligands (11, 13), which suggest that it vivo. J. Exp. Med. 185:571. may have an immunomodulatory role. Eosinophils are associated 8. Laufer, T. M., J. DeKoning, J. S. Markowitz, D. Lo, and L. H. Glimcher. 1996. with most inflammatory and infectious disorders and have been Unopposed positive selection and autoreactivity in mice expressing class II MHC only on thymic cortex. Nature 383:81. implicated beneficially in anti-tumor cytotoxicity and wound heal- 9. Delaney, J. R., Y. Sykulev, H. N. Eisen, and S. Tonegawa. 1998. Differences in ing. Yet, most of our knowledge about this cell comes from anal- the level of expression of class I major histocompatibility complex proteins on The Journal of Immunology 1975

thymic epithelial and dendritic cells influence the decision of immature thymo- 29. Luttmann, W., B. Knoechel, M. Foerster, H. Matthys, J. C. J. Virchow, and cytes between positive and negative selection. Proc. Natl. Acad. Sci. USA 95: C. Kroegel. 1996. Activation of human eosinophils by IL-13: induction of CD69 5235. surface antigen, its relationship to messenger RNA expression, and promotion of 10. Weller, P. F. 1994. Eosinophils: structure and functions. Curr. Opin. Immunol. cellular viability. J. Immunol. 157:1678. 6:85. 30. Li, L., Y. Xia, A. Nguyen, Y. H. Lai, L. Feng, T. R. Mosmann, and D. Lo. 1999. 11. Woerly, G., N. Roger, S. Loiseau, D. Dombrowicz, A. Capron, and M. Capron. Effects of Th2 cytokines on chemokine expression in the lung: IL-13 potently 1999. Expression of CD28 and CD86 by human eosinophils and role in the induces eotaxin expression by airway epithelial cells. J. Immunol. 162:2477. secretion of type 1 cytokines (interleukin 2 and interferon ␥): inhibition by im- 31. Walker, C., S. Rihs, R. K. Braun, S. Betz, and P. L. Bruijnzeel. 1993. Increased munoglobulin a complexes. J. Exp. Med. 190:487. expression of CD11b and functional changes in eosinophils after migration across 12. Anwar, A. R., R. Moqbel, G. M. Walsh, A. B. Kay, and A. J. Wardlaw. 1993. endothelial cell monolayers. J. Immunol. 150:4061. Adhesion to fibronectin prolongs eosinophil survival. J. Exp. Med. 177:839. 32. Mamalaki, C., T. Norton, Y. Tanaka, A. R. Townsend, P. Chandler, E. Simpson, 13. Lamkhioued, B., A. S. Gounni, D. Aldebert, E. Delaporte, L. Prin, A. Capron, and and D. Kioussis. 1992. Thymic depletion and peripheral activation of class I M. Capron. 1996. Synthesis of type 1 (IFN-␥) and type 2 (IL-4, IL-5, and IL-10) major histocompatibility complex-restricted T cells by soluble peptide in T-cell cytokines by human eosinophils. Ann. NY Acad. Sci. 796:203. receptor transgenic mice. Proc. Natl. Acad. Sci. USA 89:11342. 14. Weller, P. F., T. H. Rand, T. Barrett, A. Elovic, D. T. Wong, and R. W. Finberg. 33. Martin, S., and M. J. Bevan. 1997. Antigen-specific and nonspecific deletion of 1993. Accessory cell function of human eosinophils. HLA-DR-dependent, MHC- immature cortical thymocytes caused by antigen injection. Eur. J. Immunol. 27: restricted antigen-presentation and IL-1␣ expression. J. Immunol. 150:2554. 2726. 15. del Pozo, V., B. de Andre´s, E. Martıń,B.Ca´rdaba, J. C. Fernańdez, S. Gallardo, 34. Tarazona, R., O. Williams, D. Moskophidis, L. A. Smyth, Y. Tanaka, P. Tramoń, F. Leyva-Cobian, P. Palomino, and C. Lahoz. 1992. Eosinophil as M. Murdjeva, A. Wack, C. Mamalaki, and D. Kioussis. 1998. Susceptibility and antigen-presenting cell: activation of T cell clones and T cell hybridoma by eo- resistance to antigen-induced apoptosis in the thymus of transgenic mice. J. Im- sinophils after antigen processing. Eur. J. Immunol. 22:1919. munol. 160:5397. 16. Tamura, N., N. Ishii, M. Nakazawa, M. Nagoya, M. Yoshinari, T. Amano, 35. Swat, W., L. Ignatowicz, H. von Boehmer, and P. Kisielow. 1991. Clonal deletion H. Nakazima, and M. Minami. 1996. Requirement of CD80 and CD86 molecules of immature CD4ϩ8ϩ thymocytes in suspension culture by extrathymic antigen- for antigen presentation by eosinophils. Scand. J. Immunol. 44:229. presenting cells. Nature 351:150. 17. Kirberg, J., A. Baron, S. Jakob, A. Rolink, K. Karjalainen, and H. von Boehmer. 36. Owen, W. F. J., M. E. Rothenberg, D. S. Silberstein, J. C. Gasson, R. L. Stevens,

ϩ Downloaded from 1994. Thymic selection of CD8 single positive cells with a class II major his- K. F. Austen, and R. J. Soberman. 1987. Regulation of human eosinophil via- tocompatibility complex-restricted receptor. J. Exp. Med. 180:25. bility, density, and function by granulocyte/macrophage colony-stimulating fac- 18. Kisielow, P., H. Bluthmann, U. D. Staerz, M. Steinmetz, and H. von Boehmer. tor in the presence of 3T3 ﬁbroblasts. J. Exp. Med. 166:129. 1988. Tolerance in T-cell-receptor transgenic mice involves deletion of nonma- ϩ ϩ 37. Palframan, R. T., P. D. Collins, N. J. Severs, S. Rothery, T. J. Williams, and ture CD4 8 thymocytes. Nature 333:742. S. M. Rankin. 1998. Mechanisms of acute eosinophil mobilization from the bone 19. Burkly, L., C. Hession, L. Ogata, C. Reilly, L. A. Marconi, D. Olson, R. Tizard, marrow stimulated by interleukin 5: the role of speciﬁc adhesion molecules and R. Cate, and D. Lo. 1995. Expression of relB is required for the development of phosphatidylinositol 3-kinase. J. Exp. Med. 188:1621. thymic medulla and dendritic cells. Nature 373:531. 38. Rothenberg, M. E., A. D. Luster, and P. Leder. 1995. Murine eotaxin: an eosin-

20. Wack, A., H. M. Ladyman, O. Williams, K. Roderick, M. A. Ritter, and http://www.jimmunol.org/ ophil chemoattractant inducible in endothelial cells and in interleukin 4-induced D. Kioussis. 1996. Direct visualization of thymocyte apoptosis in neglect, acute tumor suppression. Proc. Natl. Acad. Sci. USA 92:8960. and steady-state negative selection. Int. Immunol. 8:1537. 39. Rothenberg, M. E., J. A. MacLean, E. Pearlman, A. D. Luster, and P. Leder. 21. Markiewicz, M. A., C. Girao, J. T. Opferman, J. L. Sun, Q. H. Hu, A. A. Agulnik, 1997. Targeted disruption of the chemokine eotaxin partially reduces antigen- C. E. Bishop, C. B. Thompson, and P. G. AshtonRickardt. 1998. Long-term T cell induced tissue eosinophilia. J. Exp. Med. 185:785. memory requires the surface expression of self- peptide major histocompatibility complex molecules. Proc. Natl. Acad. Sci. USA 95:3065. 40. Matthews, A. N., D. S. Friend, N. Zimmermann, M. N. Sarafi, A. D. Luster, 22. de Andres, B., E. Rakasz, M. Hagen, M. L. McCormik, A. L. Mueller, D. Elliot, E. Pearlman, S. E. Wert, and M. E. Rothenberg. 1998. Eotaxin is required for the A. Metwali, M. Sandor, B. E. Britigan, J. V. Weinstock, et al. 1997. Lack of Fc-⑀ baseline level of tissue eosinophils. Proc. Natl. Acad. Sci. USA 95:6273. receptors on murine eosinophils: implications for the functional significance of 41. Ardavin, C., L. Wu, C. Li, and K. Shortman. 1993. Thymic dendritic cells and T elevated IgE and eosinophils in parasitic infections. Blood 89:3826. cells develop simultaneously in the thymus from a common precursor population. 23. Witko-Sarsat, V., L. Halbwachs-Mecarelli, I. Sermet-Gaudelus, G. Bessou, Nature 362:761. 42. Douek, D. C., K. T. Corley, T. Zal, A. Mellor, P. J. Dyson, and D. M. Altmann. G. Lenoir, R. C. Allen, and B. Descamps-Latscha. 1999. Priming of blood neu- by guest on September 24, 2021 trophils in children with cystic fibrosis: correlation between functional and phe- 1996. Negative selection by endogenous antigen and superantigen occurs at mul- notypic expression of opsonin receptors before and after platelet-activating factor tiple thymic sites. Int. Immunol. 8:1413. priming. J. Infect. Dis. 179:151. 43. Marodon, G., and B. Rocha. 1994. Activation and ‘deletion’ of self-reactive 24. McGarry, M. P., and C. C. Stewart. 1991. Murine eosinophil granulocytes bind mature and immature T cells during ontogeny of Mls-1a mice: implications for the murine macrophage-monocyte specific monoclonal antibody F4/80. J. Leu- neonatal tolerance induction. Int. Immunol. 6:1899. 44. Punt, J. A., W. Havran, R. Abe, A. Sarin, and A. Singer. 1997. T cell receptor kocyte Biol. 50:471. ϩ ϩ 25. Mawhorter, S. D., D. A. Stephany, E. A. Ottesen, and T. B. Nutman. 1996. (TCR)-induced death of immature CD4 CD8 thymocytes by two distinct Identification of surface molecules associated with physiologic activation of eo- mechanisms differing in their requirement for CD28 costimulation: implications sinophils: application of whole-blood flow cytometry to eosinophils. J. Immunol. for negative selection in the thymus. J. Exp. Med. 186:1911. 156:4851. 45. Amakawa, R., A. Hakem, T. M. Kundig, T. Matsuyama, J. J. Simard, E. Timms, 26. Moqbel, R., F. Levi-Schaffer, and A. B. Kay. 1994. Cytokine generation by A. Wakeham, H. W. Mittruecker, H. Griesser, H. Takimoto, et al. 1996. Impaired eosinophils. J. Allergy Clin. Immun. 94:1183. negative selection of T cells in Hodgkin’s disease antigen CD30-deficient mice. 27. Chiaramonte, M. G., L. R. Schopf, T. Y. Neben, A. W. Cheever, Cell 84:551. D. D. Donaldson, and T. A. Wynn. 1999. IL-13 is a key regulatory cytokine for 46. Westermann, J. E., and V. E. Engelbert. 1968. Terminal variations of eosinophils Th2 cell-mediated pulmonary granuloma formation and IgE responses induced by in the rabbit thymus. Haematol. Lat. 11:199. Schistosoma mansoni eggs. J. Immunol. 162:920. 47. Lee, I., E. Yu, R. A. Good, and S. Ikehara. 1995. Presence of eosinophilic pre- 28. Zhu, Z., R. J. Homer, Z. Wang, Q. Chen, G. P. Geba, J. Wang, Y. Zhang, and cursors in the human thymus: evidence for intra-thymic differentiation of cells in J. A. Elias. 1999. Pulmonary expression of interleukin-13 causes inflammation, eosinophilic lineage. Pathol. Int. 45:655. mucus hypersecretion, subepithelial fibrosis, physiologic abnormalities, and 48. Muller, E. 1977. Localization of eosinophils in the thymus by the peroxidase eotaxin production. J. Clin. Invest. 103:779. reaction. Histochemistry 52:273.