Neonatal and Adult CD4 CD3 Cells Share Similar Gene Expression

The Journal of Immunology

Neonatal and Adult CD4؉CD3؊ Cells Share Similar Gene Expression Proﬁle, and Neonatal Cells Up-Regulate OX40 Ligand in Response to TL1A (TNFSF15)1

Mi-Yeon Kim, Kai-Michael Toellner, Andrea White, Fiona M. McConnell, Fabrina M. C. Gaspal, Sonia M. Parnell, Eric Jenkinson, Graham Anderson, and Peter J. L. Lane2

We report here the quantitative expression of a set of immunity-related genes, including TNF family members, chemokine -receptors, and transcription factors, in a CD4؉CD3؊ accessory cell. By correlating gene expression between cell-sorted popula -tions of deﬁned phenotype, we show that the genetic ﬁngerprint of these CD4؉CD3؊ cells is distinct from dendritic cells, plas macytoid dendritic cells, T cells, B cells, and NK cells. In contrast, it is highly similar to CD4؉CD3؊ cells isolated from embryonic and neonatal tissues, with the exception that only adult populations express OX40L and CD30L. We have previously reported that IL-7 signals regulate CD30L expression. In the present study, we show that both neonatal and adult CD4؉CD3؊ cells express the TNF family member, death receptor 3 (TNFRSF25), and that addition of TL1A (TNFSF15), the ligand for death receptor 3, up-regulates OX40L on neonatal CD4؉CD3؊ cells. Finally, we demonstrate that this differentiation occurs in vivo: neonatal -CD4؉CD3؊ cells up-regulate both CD30L and OX40L after adoptive transfer into an adult recipient. The Journal of Immu nology, 2006, 177: 3074–3081.

e have previously reported that CD4ϩCD3Ϫ CD30L but not OX40L expression (3). Our failure to induce CD11cϪB220Ϫ cells (CD4ϩCD3Ϫ cells) provide sur- OX40L on neonatal CD4ϩCD3Ϫ cells raised the possibility that W vival signals to activated CD4 T cells via their con- they were different cells from those that we found in adult mice. In stitutive expression of OX40L (CD252 and TNFSF4) and CD30L the present study, we report three independent pieces of evidence (CD153 and TNFSF8) in adult mouse (1, 2). These cells are that further support a relationship. We show that cells of related located in B follicles but also T cell areas, especially the outer T lineage show strong correlations in the quantitative mRNA expres- zone. In B follicles they are attached to the T cells that select sion of a 96-gene set of immunity related genes: for example, germinal center (GC)3 B cells, and in the absence of OX40 and subsets of dendritic cells (DCs), T and B cells, are closely corre- CD30 signals, GC T cells fail to survive, with consequent failure lated. This relationship also holds for neonatal and adult popula- of affinity maturation of the Ab responses (2). Even more signif- tions of CD4ϩCD3Ϫ cells. Of particular interest was the shared icantly, T cell memory for Ab responses is abrogated in OX40 and high levels of mRNA for the TNF ligands, lymphotoxin (LT) ␣, CD30 double-deficient mice (2), and we have speculated that LT␤, TNF-␣, and TNF-related activation-induced cytokine memory T cells are normally maintained by their associations with (TRANCE) (TNFSF11). Like OX40L and CD30L on adult these cells in the outer T zone. CD4ϩCD3Ϫ cells, these cells appear to express high levels of When we investigated the presence of these cells in neonatal these ligands constitutively. They also express receptor activator tissues, we found a similar population: however, OX40L and of NF-␬B (RANK) (TNFRSF11A), death receptor 3 (DR3) CD30L, the T cell survival molecules, were lacking (3). These (TNFRSF25), IL-2R␣ (CD25), IL-7R␣ (CD127), common cyto- ␥ ␥ studies demonstrated that the expression of OX40L and CD30L kine receptor -chain ( c) (CD132), CCR7, and CXCR5. was regulated independently: IL-7 signals were important for Because we found that both neonatal and adult CD4ϩCD3Ϫ cells expressed high levels of DR3, we added recombinant TL1A (TNFSF15) to both neonatal and adult populations. This signal Medical Research Council Centre for Immune Regulation, Institute for Biomedical induced high levels of OX40L expression on embryonic/neonatal Research, Birmingham Medical School, Birmingham, United Kingdom populations, and the expression of OX40L was further augmented ϩ Ϫ Received for publication April 4, 2006. Accepted for publication June 16, 2006. on adult cells. Finally, we show that embryonic CD4 CD3 cells The costs of publication of this article were defrayed in part by the payment of page following transfer into an adult recipient up-regulate OX40L and charges. This article must therefore be hereby marked advertisement in accordance CD30L expression to comparable levels to the adult host CD4ϩ with 18 U.S.C. Section 1734 solely to indicate this fact. CD3Ϫ cells in vivo. 1 This work was supported by a Wellcome Programme Grant (to P.J.L.L.). 2 Address correspondence and reprint requests to Dr. Peter J. L. Lane, Medical Re- search Council Centre for Immune Regulation, Institute for Biomedical Research, Materials and Methods Birmingham Medical School, Birmingham B15 2TT, U.K. E-mail address: Mice [email protected] All experiments were performed in accordance with the U.K. laws and with 3 ␤ ␤ Abbreviations used in this paper: GC, germinal center; 2m, 2-microglobulin; CC, Ϫ/Ϫ ␥ the approval of the local ethics committee. Normal, RAG1 , and T cell- correlation coefficient; DC, dendritic cell; DR3, death receptor 3; c, common ␥ deficient mice were bred and maintained in our animal facility. Neonatal -chain; HVEM, herpes virus entry mediator; LT, lymphotoxin; LIGHT, LT-related ϩ Ϫ inducible ligand that competes for glycoprotein D binding to HVEM on T cell; pDC, lymph node or spleen CD4 CD3 cells were isolated from 1- to 2-day-old Ϫ/Ϫ plasmacytoid DC; TRANCE, TNF-related activation-induced cytokine; RANK, re- normal BALB/c litters or RAG1 mice. Spleens from normal C57BL/6 ϩ Ϫ ceptor activator of NF-␬B. or BALB/c E15 embryos were used to isolate E15 CD4 CD3 cells.

Preparation of CD4ϩCD3Ϫ cells, plasmacytoid DCs (pDCs), DCs, and other cells Cell suspensions for isolation of CD4ϩCD3Ϫ cells, DCs, and pDCs were made from the spleens of adult RAGϪ/Ϫ mice as described previously (1, 3). Neonatal CD4ϩCD3Ϫ cells were isolated from either BALB/c or C57BL/6 mice that were 1 or 2 days old. Brieﬂy, CD11cϩ cells were positively enriched by using CD11c-coated magnetic beads (Miltenyi Biotec) and then FACS sorted into CD8ϩ and CD8Ϫ populations. CD4ϩ cells were enriched from CD11cϩ-depleted populations using CD4-coated magnetic beads, and the resulting CD4ϩ-enriched populations sorted into CD4ϩCD3ϪB220ϪCD11cϪ (CD4ϩCD3Ϫ) and CD4ϩCD3ϪB220ϩCD11clow (pDC) populations. CD45Ϫ stromal cells were FACS sorted from BALB/c mice. For the preparation of E15 CD4ϩCD3Ϫ cells, embryos from normal pregnant mice of gestation day 15 were obtained and the spleens removed. The spleens were placed in culture medium with 100 ng/ml IL-7 (Pepro- Tech) on a 0.8-␮m sterile Nuclepore ﬁlter (Millipore) on a sterile arti wrap sponge. The petri dish was then cultured in a contained humid environment

in a 10% CO2 incubator for 5 days. On day 5, cultured spleens were teased apart with fine forceps and CD4 cells enriched as above. Follicular B (CD21lowCD23ϩIgMint) cells, marginal zone B (CD21highCD23ϪIgMϩ) cells, and NK cells from normal mice were sorted to make cDNA. Th1 and Th2 cells were prepared under Th1 conditions (10 ng/ml IL-12 and 10 ␮g/ml anti-IL-4) and Th2 conditions (10 ng/ml IL-4 and 10 ␮g/ml anti-IL-12) for 6 days in vitro culture. Stimulation of E15 or neonatal or adult CD4ϩCD3Ϫ cells Prepared cells were cultured with a wide range (0.1–100 ng/ml) of recombinant mouse TL1A (R&D Systems) for 2 or 6 days of culture and then stained for flow cytometry analysis or for MoFlo cell sorting. FACS staining CD4ϩCD3Ϫ cells were stained with anti-CD4 PE, anti-CD3 FITC, anti- CD11c FITC, and anti-B220 FITC mAbs or anti-B220 allophycocyanin mAbs (BD Biosciences) and then stained with biotinylated mAbs against OX40L, CD30L, and CXCR4 (BD Biosciences) or TRANCE (R&D systems) in conjunction with streptavidin CyChrome (BD Biosciences) as the second-step staining reagent. TaqMan low-density array analysis TaqMan primer sets are designed to work with an efficiency approaching 100%, enabling the quantitative comparison of mRNA expression for different genes not only within a cell type, but also between cells of different ␤ ␤ ␤ lineages. Housekeeping genes ( -actin, 2-microglobulin ( 2m), or 18S rRNA) were used to correct for total mRNA. TaqMan low-density real-time PCR arrays (Applied Biosystems) were designed with a 96-gene format. A list of all of the genes measured is as follows: chemokines (CCL19, CXCL12, and CXCL13), chemokine receptors (CCR7, CXCR3, and CXCR5), cytokines (IL-1␣, IL-1␤, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-9, IL-10, IL-12p35, IL-12p40, IL-13, IL-15, IL-18, TSLP, IFN-␣1, IFN-␤, IFN-␥, and TGF-␤1), cytokine receptors (IL-2R␣, IL-2R␤, IL-2R␥, IL-4R␣, IL-7R␣, IL-10R␣, IL-10R␤, IL-12R␤1, IL-12R␤2, IFN-␥R1, and IFN-␥R2), costimulatory molecules (CD80, CD86, CTLA4, ICOS, and ICOSL), DC marker (DC-SIGN, ca- ␣ ␤ thepsin S, and integrin x), housekeeping (CD4, -actin, 18S RNA, and ␤ 2m), MHC class II (CD74), TLR (MyD-88, TLR2, TLR3, TLR4, TLR5,

spleens. E15 embryonic, day 2 neonatal CD4ϩCD3Ϫ cells, follicular B (fol B), marginal zone (MZ) B, and NK cells were from normal mice. Th1 and Th2 cells were differentiated in vitro for 6 days. A, Correlation of gene expressions between two cell types. Over 0.01% expression of individual mRNAs of ␤-actin signals were plotted. Axes show levels of mRNA expression relative to the ␤-actin signal (␤-actin signal ϭ 100%, log scale). Each plot compares two cell types, one on each axis (as labeled). B, CCs between cell populations. These results are representative of on average five separate experiments. C, Gene expression profile normalized to ␤-actin ␤ signals (a), to 2m(b), and to 18 rRNA (c). The left panel shows correlation of gene expressions between adult and D2 CD4ϩCD3Ϫ cells. The middle panel shows correlation of gene expressions between adult FIGURE 1. Correlation between the gene expression patterns of CD4ϩCD3Ϫ cells and CD8ϩ DCs. The right panel shows correlation of ϩ Ϫ CD4 CD3 cells and a range of other cell populations. Adult cell popu- gene expressions between adult CD4ϩCD3Ϫ cells isolated from RAG-de- ϩ Ϫ Ϫ Ϫ lations of CD4 CD3 cells, pDCs, and DCs were isolated from RAG / ficient and T cell-deficient mice. 3076 MATURATION AND FUNCTION OF ADULT CD4ϩCD3Ϫ CELLS

Table I. Expression of TNF family membersa

New TNF Adult E15 Group TNF Nomenclature Chr m/hb CD4ϩCD3Ϫ CD4ϩCD3Ϫ pDCs DCs B Cell

ILT␣ TNFSF1 17/6 ϩϩ ϩϩ Ϫ Ϫ n.d.c TNF␣ TNFSF2 17/6 ϩϩ ϩϩ Ϫ Ϫ Ϫ LT␤ TNFSF3 17/6 ϩϩ ϩϩϩ ϩ Ϫ ϩϩ LIGHT TNFSF14 17/19 ϩϩϪϪϪ TRANCE TNFSF11 14/13 ϩϩϪϪϪ

II FASL TNFSF6 1/1 ϪϪϪϪϪ OX40L TNFSF4 1/1 ϩϩϩ Ϫ Ϫ ϩ Ϫ CD30L TNFSF8 4/9 ϩϩ Ϫ Ϫ Ϫ Ϫ TL1A TNFSF15 4/9 ϪϪϪϪϪ CD27L TNFSF7 17/19 ϪϪϪϪϪ 4-1BBL TNFSF9 17/19 ϪϪϩϩϪ CD40L TNFSF5 X/X ϪϪϪϪϪ

III TWEAK TNFSF12 11/17 ϩϩϪϩϪ APRIL TNFSF13 11/17 ϪϪϪϪϪ BAFF TNFSF13B 8/13 ϪϪϪϪϪ

a ϩϩϩ, Ͼ10% ␤-actin signal; ϩϩ, 1–10%; ϩ 0.2, 1%; ϪϽ0.2%. b Chromosome number in mouse (m) and human (h). c n.d., Not determined.

TLR7, and TLR9), TNF family (LT␣, TNF-␣,LT␤, OX40L, CD40L, trophoresis and identified by fragment size using Syngene Gel Documen- FASL, CD70, CD30L, 4-1BBL, TRANCE, TWEAK, APRIL, BAFF, LT- tation Gene Tools software. related inducible ligand that competes for glycoprotein D binding to The primer sequences were as follows: ␤-actin, forward (5Ј-ATC HVEM on T cell (LIGHT), and TL1), TNFR family (TNFR1, TNFR2, TAC GAG GGC TAT GCT CTC C-3Ј) and reverse (5Ј-CTT TGA TGT LT␤R, OX40, CD40, FAS, CD30, 4-1BB, RANK, TWEAKR, BAFFR, CAC GCA CGA TTT CC-3Ј); ROR␥t, forward (5Ј-ACC TCC ACT Ј Ј HVEM, GITR, and DR3), transcription factors (Bcl-2, Bcl-6, Bcl-xL, GCC AGC TGT GTG CTG TC-3 ) and reverse (5 -CAA GTT CAG ROR␥, GATA3, foxP3, and T-bet), and others (perforin and granzyme B). GAT GCC TGG TTT CCT C-3Ј); LT␣, forward (5Ј-CTC CAT CCT cDNA was mixed with TaqMan Universal PCR Master Mix (Applied GAC CGT TGT TT-3Ј) and reverse (5Ј-TAG ACC CAC AAA AAC Biosystems). This was added to the TaqMan Low-Density Array, and PCR CCT GC-3Ј); TL1A, forward (5Ј-AGTCCCAGTGGAAGTGCTG-3Ј) was performed in a 7900HT Fast Real-Time PCR System (Applied Bio- and reverse (5Ј-GTGCTAAGTCCTGCGAGGAT-3Ј). systems) according to manufacturer’s recommendations. The relative signal per cell was quantified by setting a threshold within the logarithmic phase of the PCR and determining the cycle number at Results ϩ Ϫ which the fluorescence signal reached the threshold (Ct). The Ct for the Embryonic/neonatal and adult CD4 CD3 cells display similar ␤ target gene was subtracted from the Ct for -actin. The relative amount was genetic fingerprints that readily distinguish them from other Ϫ⌬Ct ϫ 2 calculated as 2 10 . cell types Conventional PCR analysis Embryonic/neonatal CD4ϩCD3Ϫ cells share a common surface Signals for ROR␥t, LT␣, TL1A, and ␤-actin were analyzed by conven- phenotype with the cells that we have described in adult mice: ϩ Ϫ Ϫ Ϫ ␣ϩ ␥ ϩ ϩ tional PCR. PCR products were analyzed by ethidium bromide gel elec- CD4 CD3 CD11c B220 IL-7R , cytokine c , CD45 ,

Table II. Expression of TNFR family membersa

New TNFR Adult E15 Group TNFR Nomenclature Chr m/hb CD4ϩCD3Ϫ CD4ϩCD3Ϫ pDCs DCs B Cell

I RANK TNFRSF11A 1/18 ϩϩ ϩϩ Ϫ ϩϩ Ϫ TNFR1 TNFRSF1A 6/12 ϩϩϩϩϩϩϪ LT␤R TNFRSF3 6/12 ϩϩϪϪϪ

II CD27c TNFRSF7 6/12 ϪϪϪϪn.d. CD40 TNFRSF5 2/20 ϩ/Ϫϩ/Ϫ Ϫ ϩϩ ϩϩ OX40 TNFRSF4 4/1 ϪϪϪϩϩ/Ϫ CD30 TNFRSF8 4/1 ϪϪϪϪϪ GITR TNFRSF18 4/1 ϪϪϪϪϩ/Ϫ TNFR2 TNFRSF1B 4/1 ϩϩ ϩϩ ϩ ϩϩ ϩ/Ϫ 4-1BB TNFRSF9 4/1 ϩϩϪϩϪ HVEM TNFRSF14 4/1 ϩϩ ϩϩ ϩϩ ϩ Ϫ DR3 TNFRSF25 4/1 ϩϩ ϩϩ Ϫ Ϫ Ϫ

III TWEAKR TNFRSF12 17/16 ϪϪϪϪϪ BAFFR TNFRSF13C 15/22 ϪϪϪϪϩϩ

a ϩϩϩ, Ͼ10% ␤-actin signal; ϩϩ, 1–10%; ϩ 0.2, 1%; ϪϽ0.2%. b Chromosome number in mouse (m) and human (h). c Detected by ﬂow cytometry. d n.d., Not determined. The Journal of Immunology 3077

weaker CCs. In contrast, after OX40L and CD30L gene expression was excluded from the analysis, the gene expression in adult CD4ϩCD3Ϫ cells was strongly correlated with embryonic (E15) spleen CD4ϩCD3Ϫ cells (CC ϭ 0.86) and with neonatal (D2) spleen CD4ϩCD3Ϫ cells (CC ϭ 0.90); neonatal lymph node and neonatal spleen CD4ϩCD3Ϫ cells were also strongly correlated (CC ϭ 0.88) (Fig. 1, A and B). The values of individual gene expression are shown in supplemental Fig. 1 data.4 The correlation between genes expressed in individual cell types is independent of correction for total mRNA expression as it de- pends on the relative ranking of individual genes that are analyzed. Correction for housekeeping genes, however, is useful as it allows comparison of levels of expression of individual mRNAs. If levels of expression of mRNA after correction for housekeeping genes are similar, the slope of a line drawn through the correlated genes will be one. As indicated in Fig. 1C, not only are the genes correlated for neonatal and adult CD4ϩCD3Ϫ populations, but the slope of a line drawn through the points representing individual mRNAs is ϳ ␤ ␤ 1 whether -actin (Fig. 1Ca), 2m (Fig. 1Cb), or 18S rRNA (Fig. 1Cc) is used as the housekeeping gene for correction. RAGϪ/Ϫ or T cell-deficient mice were used as a source of adult CD4ϩCD3Ϫ cells because of the technical difficulties in purifying CD4ϩCD3Ϫ cells from mice with an intact repertoire of CD4 T cells, which adhere to and contaminate the CD4ϩCD3Ϫ population (1). We have previously reported that ϳ105 CD4ϩCD3Ϫ cells can be isolated from an individual RAGϪ/Ϫ spleen and more from T cell-deficient mice (3). We think it likely, however, that this un- derestimates their number as they attach to VCAM-1ϩ stromal cell populations in both B and T cell areas, making their isolation difficult (our unpublished observations). By confocal microscopy, we think that there are as many CD4ϩCD3Ϫ cells as DCs, but the latter are much more readily isolated from tissues (1). Cross-correlation of gene expression between CD4ϩCD3Ϫ cells from adult RAGϪ/Ϫ and T cell-deficient mice that have normal numbers of B cells (CC ϭ 0.92) (Fig. 1C) indicates that the pattern of gene expression does not depend on B cells. Furthermore, the slope of the correlated genes is ϳ1, indicating that levels of genes FIGURE 2. Expression of Bcl family members and chemokines/chemo- expressed are not influenced significantly by B cells. kine receptors. A, Relative mRNA expression of Bcl-2, Bcl-xL, CCR7 and ϩ Ϫ its ligand (CCL19), CXCR5 and its ligand (CXCL13), and CXCR3 nor- Expression of TNF/TNFR family members in CD4 CD3 cells malized to ␤-actin signals. This result is representative of three separate There are currently ϳ17 identified TNF and ϳ30 TNFR family experiments. B, Relative mRNA expression normalized to ␤ m signals. C, 2 members (͗www.gene.ucl.ac.uk/nomenclature/genefamily/tnftop. Surface CXCR4 expression on neonatal D2 and adult CD4ϩCD3Ϫ cells. html͘). Our gene array focused on TNF/TNFR family members Shaded histograms show control staining with biotinylated rat Abs. linked with survival. The details of the distinctive gene profile of TNF and TNFR (TNF/TNFR) family members established for the ϩ Ϫ ϩ ϩ ϩ low CD4 CD3 cell type are tabulated (Tables I and II). Expression of Thy1 , TRANCE , RANK , and MHC class II (4). Although mRNA in embryonic E15 CD4ϩCD3Ϫ cells (Tables I and II) was relatively few genes are cell specific, we reasoned that if we com- similar to D2 neonatal CD4ϩCD3Ϫ cells (data not shown). Our pared the levels of mRNA expression for a more comprehensive set analysis focused on genes expressed at high levels (mRNA ex- of immunity-related genes, cells of related lineage would share a pressed at Ͼ0.2% of the ␤-actin signal). To simplify analysis, gene genetic fingerprint, particularly if we looked at genes linked with expression has been categorized into four groups relative to ex- immune function and migration. Therefore, we designed TaqMan pression of ␤-actin: 1) ϩϩϩ Ͼ10%; 2) ϩϩ 1–10%; 3) ϩ 0.2, 1%; arrays for a panel of immunity-related genes (see Materials and and 4) ϪϽ0.2%. Methods). Comparison of genes expressed within the two major ϩ ϩ Ϫ ϩ TNF family members associated with B cell survival and acti- subsets of DCs in mice, CD8 CD11c and CD8 CD11c DCs vation (APRIL, BAFF, and CD40L) do not appear in the gene ϭ (correlation coefficient (CC) 0.94), Th1- and Th2-differentiated profiles of any of the CD4ϩCD3Ϫ cell populations. Of the 15 TNF ϭ T cells (CC 0.86), and marginal zone and follicular B cells family members on our array (Tables I and II), 8 were expressed ϭ (CC 0.95), revealed a strong CC for the related cells (Fig. 1A, in mRNA from adult CD4ϩCD3Ϫ cells. OX40L and CD30L are top panel), but there was little correlation between cells of different selectively expressed on adult cells, and we have linked them with types, lending validity to the use of the fingerprinting method. ϩ Ϫ T cell survival for B cell help (2); otherwise embryonic and neo- Comparison of gene expression between adult CD4 CD3 cells natal populations expressed a similar pattern. A striking feature of and either NK (CC ϭ 0.66), pDCs (CC ϭ 0.59), CD8ϩ DCs (CC ϭ 0.68), follicular B (CC ϭ 0.70), marginal zone B cells (CC ϭ 0.65), Th2 (CC ϭ 0.63), or Th1 (CC ϭ 0.76) showed much 4 The online version of this article contains supplemental material. 3078 MATURATION AND FUNCTION OF ADULT CD4ϩCD3Ϫ CELLS this pattern is the presence of the ligands for TNFR1 (TNF-␣ and ROR␥t mRNA is expressed in adult CD4ϩCD3Ϫ cells, and DR3 LT␣) and LT␤R (LT␣ and LT␤ and LIGHT), all of which are signals up-regulate OX40L expression on embryonic/neonatal linked with the segregation of B:T areas in lymphoid tissue (5); the CD4ϩCD3Ϫ cells TNFR1 ligands and LT␤R ligands with TRANCE are linked with Because of the similar genetic fingerprint of embryonic/neonatal lymph node organogenesis (6). ϩ Ϫ and adult CD4 CD3 cells, we looked for mRNA expression of Neonatal and adult CD4ϩCD3Ϫ cells coordinately expressed 7 the splice variant of the retinoic acid orphan receptor, ROR␥t (12), of the 14 TNFR family members, 5 of them strongly (ϩϩ). The ϩ Ϫ a gene essential for the function of CD4 CD3 cells in lymph TNFR family members can also be grouped into those involved in node development. Both adult CD4ϩCD3Ϫ cells and embryonic/ lymph node development and B:T segregation (5, 7) (LT␤R, neonatal CD4ϩCD3Ϫ cells expressed mRNA for ROR␥t, but levels TNFR1, and RANK) and those linked with T cell activation (HVEM, were 4-fold higher in embryonic CD4ϩCD3Ϫ cells than in adult TNFR2, 4-1BB, and DR3) (8). These four T cell-associated TNFR ϩ Ϫ A family members come from a gene cluster of seven TNFR family CD4 CD3 cells (Fig. 3 ) and both expressed both TNFR1 li- ␤ members on human chromosome 1 and mouse chromosome 4. Nei- gands and LT R ligands (Fig. 3A and Tables I and II). ther neonatal nor adult CD4ϩCD3Ϫ cells express CD30 or OX40. The key difference between embryonic/neonatal and adult CD4ϩCD3Ϫ cells is that the former lack expression of the T cell Expression of non-TNF/TNFR family genes survival TNF ligands, OX40L and CD30L (Fig. 3A) (3). We have previously found that IL-7 signals up-regulate CD30L expression Fig. 2 summarizes the expression of non-TNF/TNFR family genes ϩ Ϫ expressed on CD4ϩCD3Ϫ cell populations normalized to ␤-actin on neonatal CD4 CD3 cells, but these signals had no effect on ␤ OX40L expression. Our TaqMan low-density arrays had demon- (Fig. 2A)or 2m (Fig. 2B). All of them were expressed at com- strated mRNA for the TNFR family member, DR3 (TNFRSF25), parable levels in adult and embryonic/neonatal populations inde- ϩ Ϫ pendently on housekeeping genes. There are three gene groups of on both embryonic/neonatal and adult CD4 CD3 cells (Tables I and II), so we tested the effects of the recombinant ligand, TL1A particular interest: the chemokine receptors, the survival genes, ϩ Ϫ and the GC-specific genes. CD4ϩCD3Ϫ cells express both of the (TNFSF15), on embryonic, neonatal, and adult CD4 CD3 cells. chemokine receptors, CXCR5 and CCR7, but not the pDC-related TL1A was added at 100 ng/ml for 2 days in culture (Fig. 3, B and receptor, CXCR3 (9). We looked for but did not find mRNA for the C), and similar results were obtained with TL1A added at 1 ng/ml ϩ Ϫ ligands of CXCR5 (CXCL13) and CCR7 (CCL21 and CCL19), (data not shown). In adult CD4 CD3 cells, TL1A down-regu- which occur in stromal populations (10, 11). Because TaqMan primer lated mRNA expression for ROR␥t and up-regulated the expression for CXCR4 was not available, we stained with mAbs and identified of OX40L and TRANCE (Fig. 3B). ϩ Ϫ expression of CXCR4 on both neonatal and adult CD4ϩCD3Ϫ cells Addition of TL1A to embryonic (E15) CD4 CD3 cells also (Fig. 2C). down-regulated ROR␥t expression with up-regulation of both

The survival genes, Bcl-2 and Bcl-xL, appear at high levels in all TRANCE and particularly OX40L (Fig. 3C). The effects on mRNA CD4ϩCD3Ϫ populations (Fig. 2, A and B). This helps explain our were reﬂected by changes in protein expression at the cell surface observation that CD4ϩCD3Ϫ cells survive in culture for at least a on adult splenic CD4ϩCD3Ϫ cells (Fig. 4A), adult lymph node week (our unpublished observations). CD4ϩCD3Ϫ cells (Fig. 4B), neonatal (D2) splenic CD4ϩCD3Ϫ

FIGURE 3. Effects of TL1A signals on CD4ϩCD3Ϫ cells. A, Genes differ- entially expressed in embryonic (E15) and adult CD4ϩCD3Ϫ cells. B, Adult CD4ϩCD3Ϫ cells cultured with or without 100 ng/ml TL1A for 2 days. C, E15 CD4ϩCD3Ϫ cells cultured with or without 100 ng/ml TL1A for 2 days. D, Correlation of gene expression in OX40Lϩ vs OX40LϪ FACS-sorted E15 CD4ϩCD3Ϫ cells after treatment with TL1A for 2 days. Axes show levels of mRNA expression relative to the ␤-actin signal (␤-actin signal ϭ 100%, log scale). These results are representative of at least two separate experiments. The Journal of Immunology 3079

FIGURE 4. Effect of TL1A on TNF ligand protein expression on CD4ϩCD3Ϫ cells in vitro. A, Adult splenic CD4ϩCD3Ϫ cells cultured with/without 100 ng/ml TL1A for 2 days. B, Adult lymph node CD4ϩ CD3Ϫ cells cultured with/without 100 ng/ml TL1A for 2 days. C, Neonatal splenic day 2 CD4ϩCD3Ϫ cells cultured with/without 100 ng/ml TL1A for 2 days. D, Embryonic splenic E15 CD4ϩCD3Ϫ cells cultured with/without 100 ng/ml TL1A for 2 days. E, Neonatal splenic day 2 CD4ϩCD3Ϫ cells cultured with/without 100 ng/ml TL1A and/or 100 ng/ml IL-7 for 6 days. Shaded histograms show control staining with biotinylated rat Abs. This result is representative of four separate experiments.

cells (Fig. 4C), and E15 splenic CD4ϩCD3Ϫ cells (Fig. 4D). In all defined by their expression of CD4 and absence of CD11c, B220, three groups, 48 h of DR3 signals up-regulated OX40L and and CD3: one a CD4ϩCD3Ϫ cell, the other the precursor of the TRANCE expression but had little effect on CD30L expression. OX40LϩCD4ϩCD3Ϫ adult phenotype cell. To examine this pos- ϩ Ϫ Costimulation of neonatal CD4 CD3 cells with TL1A for 6 days sibility, we compared the genetic fingerprint of DR3-signaled cells up-regulated OX40L, TRANCE, and also CD30L, whereas IL-7 that were OX40Lϩ or OX40LϪ. With the exception of OX40L, the alone up-regulated CD30L and TRANCE but not OX40L (Fig. gene profiles were highly correlated (CC ϭ 0.95) (Fig. 3D), sug- 4E). Together, IL-7 and TL1A showed additive effects. However, gesting a single population of cells. We think the reason that not all the fact that mice deficient in ␥ or IL-7 signals have normal levels ϩ Ϫ c CD4 CD3 cells up-regulate OX40L is technical and related to of OX40L but not CD30L and TRANCE shows an important role the diffusion of TL1A into the embryonic spleen fragments. In for IL-7 signals in CD30L expression (3), and therefore, the effects of TL1A in the 6 day experiments may be indirectly mediated pilot experiments, when TL1A was added to whole embryonic through IL-7. spleens, OX40L induction was only seen on a small fraction of the ϩ Ϫ Because only ϳ60% of E15 CD4ϩCD3Ϫ cells expressed high CD4 CD3 cells; this fraction increased substantially to the levels levels of OX40L after the addition of TL1A (Fig. 4D), it was reported when the cultured E15 spleens were teased apart (see possible that there were two precursors within the population Materials and Methods). 3080 MATURATION AND FUNCTION OF ADULT CD4ϩCD3Ϫ CELLS

FIGURE 5. In vivo up-regulation of OX40L and CD30L on embryonic CD4ϩCD3Ϫ cells (CD45.2) after transfer into an adult mouse (CD45.1). Five days after transfer of embryonic (E15) CD4ϩCD3Ϫ cells into an adult mouse, the host spleen was taken, and the CD11c-depleted CD4-enriched population was overnight cultured and immu- nostained the following day. Shaded histograms show control staining with biotinylated rat Abs. This result is representative of two separate experiments.

Fetal CD4ϩCD3Ϫ cells up-regulate expression of both OX40L E15 spleen expresses TL1A mRNA and CD30L after transfer into adult recipients Because TL1A protein induced OX40L expression in embryonic To test directly whether embryonic CD4ϩCD3Ϫ cells were capa- CD4ϩCD3Ϫ cells, we hypothesized that TL1A expression would ble of up-regulating OX40L and CD30L in vivo, CD4ϩCD3Ϫ cells be minimal in E15 spleen. Due to lack of reagents to detect TL1A were prepared from CD45.2 embryonic spleens and transferred protein, we tested for mRNA expression from total mRNA isolated into an adult CD45.1 recipient that lacked T cells (13) (isolation of from E15, D1, and adult spleen (Fig. 6). TL1A mRNA was clearly CD4ϩCD3Ϫ populations from T cell-sufficient mice is technically expressed in E15 tissues but not E15 CD4ϩCD3Ϫ cells. We do not difficult (3)). Five days later, CD4ϩ cells were enriched from the know, however, whether TL1A is expressed at the protein level, spleen and stained with the allotype marker to identify transferred but it is clearly not able to signal through DR3 expressed on embryonic CD4ϩCD3Ϫ cells and OX40L and CD30L (Fig. 5). CD4ϩCD3Ϫ cells. Transferred CD4ϩCD3Ϫ cells were clearly identifiable in adoptive recipients, and while they were negative for CD30L and OX40L Discussion before cell transfer, they showed expression levels of OX40L and We have previously reported that by virtue of its constitutive ex- CD30L comparable to host adult CD4ϩCD3Ϫ cells, indicating that pression of the TNF family members, OX40L and CD30L (1, 2, 4), ϩ Ϫ fetal CD4ϩCD3Ϫ cells acquire hallmarks of adult CD4ϩCD3Ϫ the adult CD4 CD3 cell (present in all secondary lymphoid tis- cells in vivo. sues that we have examined) plays a critical role both in main- taining the T cells that select B cells within GCs and in forming the memory T cells that provide help for secondary B cell responses. We used a set of 96 immunity-related genes to identify a relationship between the adult CD4ϩCD3Ϫ cells and other cell populations. In the present study, we demonstrate that these cells share a common phenotype with CD4ϩCD3Ϫ cells found in embryo and neonate, including TNF family members (LT␣,LT␤, TNF-␣, TRANCE, and LIGHT), cytokine receptors (IL-2R␣, IL-7R␣, and ␥ c), chemokine receptors (CCR7 and CXCR5) (allowing their lo- calization in B and T cell areas), and survival molecules (bcl-2 and

bcl-xL). Levels of expression are comparable to those expressed in ϩ Ϫ FIGURE 6. TL1A mRNA expression in embryonic, neonatal, and adult embryonic and neonatal CD4 CD3 cells and at least an order of ϩ spleens. cDNA was prepared from embryonic E15, neonatal D1, and adult magnitude greater than in CD11c DCs or pDCs. wild-type mice and E15 CD4ϩCD3Ϫ cells. Expression of TL1A and ␤-ac- Although embryonic/neonatal and adult CD4ϩCD3Ϫ cells share tin was assessed by PCR. Legend to supplemental data Fig. 1.4 The values a similar genotype compared with other cell types and also share ␤ ␤ of individual mRNA expression normalized to the 2m signal ( 2m sig- expression of similar set of protein markers at the cell surface, they ϭ nal 100). Each plate contains four samples derived from FACS-sorted clearly differ in their expression of the T cell survival proteins, cell populations. Analysis from four chips is shown in this ﬁgure. PCR was CD30L and OX40L, which may help explain why T cell priming done for 40 cycles. The signal for each gene is the cycle number at which the ﬂuorescence signal reached the threshold (Ct). Ct was subtracted from in the neonate results in tolerance rather than autoimmunity (3). ␤ Ϫ ␤ We have previously reported that CD30L expression can be in- the signal for housekeeping gene, in this case 2m (Ct Ct 2m). The dif- ϩ Ϫ ference between each gene and the housekeeping gene was calculated as duced on neonatal CD4 CD3 cells in vitro with IL-7, and the 2Ϫ(CtϪCt␤2m) ϫ 102. This value gives individual mRNA expression nor- expression of CD30L in vivo also appears to be IL-7 dependent ␤ ␤ malized to the 2m signal. The 2m signal is 100. (3). In the present study, we have shown that a signal from another The Journal of Immunology 3081

TNF family member, TL1A, through DR3 expressed on both em- References ϩ Ϫ bryonic/neonatal and adult CD4 CD3 cells rapidly up-regulates 1. Kim, M. Y., F. M. Gaspal, H. E. Wiggett, F. M. McConnell, A. Gulbranson-Judge, C. Raykundalia, L. S. Walker, M. D. Goodall, and the expression of the T cell survival ligand, OX40L, on embryonic/ ϩ Ϫ ϩ Ϫ P. J. Lane. 2003. CD4 CD3 accessory cells costimulate primed CD4 T cells neonatal CD4 CD3 cells and induces further up-regulation of Immunity ϩ Ϫ through OX40 and CD30 at sites where T cells collaborate with B cells. OX40L on adult CD4 CD3 cells. 18: 643–654. TL1A has been reported to be inducible by TNF-␣ and IL-1␣ 2. Gaspal, F. M., M. Y. Kim, F. M. McConnell, C. Raykundalia, V. Bekiaris, and P. J. Lane. 2005. Mice deficient in OX40 and CD30 signals lack memory anti- (14) and thought to circulate as a homotrimeric soluble form (15). body responses because of deficient CD4 T cell memory. J. Immunol. 174: TL1A is produced by cerebral endothelial cells in mouse (16) and 3891–3896. umbilical vein endothelial cells in human (17) and inhibits angio- 3. Kim, M.-Y., G. Anderson, I.-L. Martensson, L. Erlandsson, W. Arlt, A. White, and P. J. L. Lane. 2005. OX40-ligand and CD30-ligand are expressed on adult but genesis (18). In addition, membrane TL1A is expressed by human not neonatal CD4ϩCD3Ϫ inducer cells: evidence that IL7 signals regulate CD30- CCR9ϩ mucosal and gut-homing peripheral blood T cells (19). ligand but not OX40-ligand expression. J. Immunol. 174: 6686–6691. 4. Lane, P. J. L., M. C. Gaspal, and M.-Y. Kim. 2005. Two sides of a cellular coin: DR3 is the receptor for TL1A containing a death domain with the CD4ϩCD3Ϫ cells orchestrate memory antibody responses and lymph node or- highest homology to TNFR1 (20, 21) and, in addition to being ganisation. Nat. Rev. Immunol. 5: 655–660. expressed on CD4ϩCD3Ϫ cells, is expressed by activated T cells 5. Fu, Y. X., and D. D. Chaplin. 1999. Development and maturation of secondary lymphoid tissues. Annu. Rev. Immunol. 17: 399–433. (15). Although DR3 contains a death domain, DR3/TL1A engage- 6. Kim, D., R. E. Mebius, J. D. MacMicking, S. Jung, T. Cupedo, Y. Castellanos, ment also leads to activation of NF-␬B and survival rather than J. Rho, B. R. Wong, R. Josien, N. Kim, et al. 2000. Regulation of peripheral apoptosis (14, 22). Our data demonstrating normal expression of lymph node genesis by the tumor necrosis factor family member TRANCE. ϩ Ϫ J. Exp. Med. 192: 1467–1478. OX40L on CD4 CD3 cells in T cell-deficient mice (3) suggest 7. Mebius, R. E. 2003. Organogenesis of lymphoid tissues. Nat. Rev. Immunol. 3: that T cell-derived TL1A is not essential for expression of OX40L 292–303. ϩ Ϫ 8. Croft, M. 2003. Co-stimulatory members of the TNFR family: keys to effective on CD4 CD3 cells. Nat. Rev. Immunol. ϩ Ϫ T cell immunity? 3: 609–620. However, neonatal CD4 CD3 cells are found clustered close 9. Cella, M., D. Jarrossay, F. Facchetti, O. Alebardi, H. Nakajima, A. Lanzavecchia, to blood vessels in spleen (our unpublished data), which could and M. Colonna. 1999. Plasmacytoid monocytes migrate to inflamed lymph ␣ nodes and produce large amounts of type I interferon. Nat. Med. 5: 919–923. therefore potentially provide these DR3 signals (14). TNF- ex- 10. Gunn, M. D., V. N. Ngo, K. M. Ansel, E. H. Ekland, J. G. Cyster, and ϩ Ϫ pressed by neonatal CD4 CD3 cells has the potential to induce L. T. Williams. 1998. A B-homing chemokine made in lymphoid follicles acti- TL1A (14) and also IL-7 (23) expression on endothelium, provid- vates Burkitt’s lymphoma type receptor-1. Nature 391: 799–802. 11. Luther, S. A., H. L. Tang, P. L. Hyman, A. G. Farr, and J. G. Cyster. 2000. ing a signaling loop leading to their own maturation to OX40L and Coexpression of the chemokines ELC and SLC by T zone stromal cells and CD30L expressing adult phenotype CD4ϩCD3Ϫ cells. This con- deletion of the ELC gene in the plt/plt mouse. Proc. Natl. Acad. Sci. USA 97: clusion is further supported by our observations that neonatal 12694–12699. ϩ Ϫ 12. Sun, Z., D. Unutmaz, Y. R. Zou, M. J. Sunshine, A. Pierani, S. Brenner-Morton, CD4 CD3 cells up-regulate expression of both OX40L and R. E. Mebius, and D. R. Littman. 2000. Requirement for ROR␥ in thymocyte CD30L to levels comparable to that found on adult cells following survival and lymphoid organ development. Science 288: 2369–2373. ϩ Ϫ 13. Hollander, G. A., B. Wang, A. Nichogiannopoulou, P. P. Platenburg, adoptive transfer of embryonic CD4 CD3 cells into adult mice W. van Ewijk, S. J. Burakoff, J. C. Gutierrez-Ramos, and C. Terhorst. 1995. in vivo. Because TL1A mRNA is detected in E15 spleen, lack of Developmental control point in induction of thymic cortex regulated by a sub- TL1A mRNA could not explain the failure of embryonic population of prothymocytes. Nature 373: 350–353. ϩ Ϫ 14. Migone, T. S., J. Zhang, X. Luo, L. Zhuang, C. Chen, B. Hu, J. S. Hong, CD4 CD3 cells to express OX40L. One possibility is that TL1A J. W. Perry, S. F. Chen, J. X. Zhou, et al. 2002. TL1A is a TNF-like ligand for protein is not present in E15 tissues; alternatively, it may be se- DR3 and TR6/DcR3 and functions as a T cell costimulator. Immunity 16: 479–492. questered inside cells and only available in response to some as yet 15. Kim, S., and L. Zhang. 2005. Identification of naturally secreted soluble form of unidentified signal present in postnatal life. This is currently under TL1A, a TNF-like cytokine. J. Immunol. Methods 298: 1–8. investigation. 16. Xiao, Q., C. Y. Hsu, H. Chen, X. Ma, J. Xu, and J. M. Lee. 2005. Characterization ϩ Ϫ of cis-regulatory elements of the vascular endothelial growth inhibitor gene pro- Both embryonic/neonatal and adult CD4 CD3 cells express a moter. Biochem. J. 388: 913–920. second set of TNF family members, including LT␣,LT␤, TNF-␣, 17. Prehn, J. L., S. Mehdizadeh, C. J. Landers, X. Luo, S. C. Cha, P. Wei, and S. R. Targan. 2004. Potential role for TL1A, the new TNF-family member and and LIGHT, that are linked with the organized B:T segregation ␥ ϩ Ϫ potent costimulator of IFN- , in mucosal inflammation. Clin. Immunol. 112: observed in lymphoid tissues (5). These CD4 CD3 cells have a 66–77. similar phenotype to lymphoid tissue inducer cells, which, via their 18. Zhai, Y., J. Ni, G. W. Jiang, J. Lu, L. Xing, C. Lincoln, K. C. Carter, F. Janat, ␣ ␤ ␣ D. Kozak, S. Xu, et al. 1999. VEGI, a novel cytokine of the tumor necrosis factor expression of LT ,LT , TNF- , and LIGHT, elicit the develop- family, is an angiogenesis inhibitor that suppresses the growth of colon carcino- ment of lymph nodes and gut-associated lymphoid tissue (7) and mas in vivo. FASEB J. 13: 181–189. have been associated with the initial B:T segregation that occurs in 19. Papadakis, K. A., D. Zhu, J. L. Prehn, C. Landers, A. Avanesyan, G. Lafkas, and S. R. Targan. 2005. Dominant role for TL1A/DR3 pathway in IL-12 plus IL-18- the neonatal lymph node (24). Although definitive proof that lym- induced IFN-␥ production by peripheral blood and mucosal CCR9ϩ T lympho- ϩ Ϫ phoid tissue inducer cells give rise to adult CD4 CD3 cells cytes. J. Immunol. 174: 4985–4990. awaits cell fate mapping experiments (25), we think the data pro- 20. Kitson, J., T. Raven, Y. P. Jiang, D. V. Goeddel, K. M. Giles, K. T. Pun, C. J. Grinham, R. Brown, and S. N. Farrow. 1996. A death-domain-containing vided here are suggestive of this sequence of events. receptor that mediates apoptosis. Nature 384: 372–375. In summary, in this report, we provide evidence that the pre- 21. Chinnaiyan, A. M., K. O’Rourke, G. L. Yu, R. H. Lyons, M. Garg, D. R. Duan, ϩ Ϫ L. Xing, R. Gentz, J. Ni, and V. M. Dixit. 1996. Signal transduction by DR3, a cursors of adult CD4 CD3 cells are present in embryonic tissues death domain-containing receptor related to TNFR-1 and CD95. Science 274: but lack expression of the TNF ligands that we have linked with T 990–992. cell memory, OX40L and CD30L. We demonstrate, however, that 22. Wen, L., L. Zhuang, X. Luo, and P. Wei. 2003. TL1A-induced NF-␬B activation and c-IAP2 production prevent DR3-mediated apoptosis in TF-1 cells. J. Biol. the embryonic population differentiates into adult phenotype cells Chem. 278: 39251–39258. following adoptive transfer in vivo in adult hosts. Furthermore, we 23. Kroncke, R., H. Loppnow, H. D. Flad, and J. Gerdes. 1996. Human follicular demonstrate that TL1A, the ligand for the TNFR, DR3, expressed dendritic cells and vascular cells produce interleukin-7: a potential role for interleukin-7 in the germinal center reaction. Eur. J. Immunol. 26: 2541–2544. on both embryonic/neonatal and adult populations, rapidly up-reg- 24. Cupedo, T., F. E. Lund, V. N. Ngo, T. D. Randall, W. Jansen, M. J. Greuter, ulates OX40L expression on the populations. R. de Waal-Malefyt, G. Kraal, J. G. Cyster, and R. E. Mebius. 2004. Initiation of cellular organization in lymph nodes is regulated by non-B cell-derived signals and is not dependent on CXC chemokine ligand 13. J. Immunol. 173: 4889–4896. Disclosures 25. Eberl, G., and D. R. Littman. 2004. Thymic origin of intestinal ␣␤ T cells re- The authors have no financial conflict of interest. vealed by fate mapping of ROR␥tϩ Cells. Science 305: 248–251.