Positive Selection by Purified MHC Class II / Thymic Epithelial Cells In Vitro: Costimulatory Signals Mediated by B7 Are Not Involved

We have investigated the possibility that the costimulatory signals required for activation of mature T cells also play a role in providing differentiation signals for positive selection during T-cell development. We show that purified MHC Class II+ thymic epithelial cells are able to support positive selection in vitro but lack both the functional capacity to deliver costimulatory signals and expression of the costimulatory ligand B7. Our results suggest that the additional signals provided by costimulatory ligands are not required for TCR-mediated positive selection, although other ancillary signals provided by thymic epithelial cells may be involved.

INTRODUCTION T-cell precursors entering the thymus undergo a program of proliferation and differentiation to produce a population of CD4+8 + cortical thymocytes expressing low levels of the T-cell receptor (TCR). Further maturation into single positive CD4 or CD8 cells requires positive selection and is triggered by signaling through the TCR in those cells recognizing major histocompatibility complex (MHC) molecules on the thymic epithelium (reviewed in von Boehmer, 1992). However, engaging the TCR on CD4 / 8 cells can also lead to negative selection, by the induction of apoptosis or programmed cell death (Smith et al., 1989). Thus, TCR-mediated signaling can have at least two different consequences in CD4+8 cells, suggesting that additional factors may be involved in regulating the consequences of TCR engagement at this stage of development.
In mature T cells, activation and proliferation in response to antigen requires dual signaling involving both TCR ligation and costimulatory signals mediated by interactions between CD28 on T cells "Corresponding author. and B7 on antigen-presenting cells (reviewed in Lui and Linsley, 1992;Jenkins and Johnson, 1993). In contrast, CD28-B7 mediated costimulation does not appear to be required for negative selection (Tan et al., 1992;Jones et al., 1993). However, a possible role for costimulatory signals in positive selection has been raised by the finding of a number of similarities between this process and activation (Bendelac et al., 1992).
Recently, we have developed a reaggregate organ culture system that supports the differentiation of CD4 / 8 cells in vitro when they are combined with selected thymic stromal cells (Jenkinson et al., 1992). This has enabled us to address directly the role of costimulatory signals in positive selection by examining the ability of thymic epithelium to provide these signals under conditions where it is competent to support positive selection. We show that purified MHC Class II epithelial cells can supply all the signals required for positive selection but lack both the ability to provide functional costimulatory signals and expression of the costimulatory ligand B7. These results provide direct evidence that the costimulatory signals involved in the activation of mature T cells are not required for the positive selection and consequent differentiation of CD4 8 thymocytes. We have previously shown that dGuo treated fetal thymus lobes consist of a mixture of cortical and medullary epithelial cells, fibroblasts, and macrophages, and contain all the components required to support the maturation of CD4 8 + thymocytes into functionally competent single positive CD4 / or CD8 cells (Jenkinson et al., 1992). To determine whether purified MHC Class II / cortical epithelial cells can act as the sole source of signals required for maturation, we looked at their ability to support the generation of TCR / single positive CD4 / or CD8 cells from TCR-CD4 / 8 precursors in reaggregate cultures. This approach excludes the involvement of other stromal cell types as source of additional differentiation signals, unlike studies where thymic epithelial cell lines (Hugo et al., 1992;Vukmanovic et al., 1992) or bone marrow-derived cells (Bix and Raulet, 1992) have been shown to mediate positive selection when introduced into the thymus in vivo where endogenous stromal components are still present.
As shown in Fig. 1, purified MHC Class II epithelial cells support both the phenotypic and functional maturation of CD4 / 8 cells, resulting in the appearance of CD4 or CD8 single positive cells with upregulated levels of TCR and the ability (C) This conclusion is further supported by the ability of cells recovered from these cultures to make a specific proliferative response when stimulated with the V,88 specific superantigen SEB, indicating that functional as well as phenotypic maturation has taken place. imply that thymic epithelial cells should be able to provide them. Previous attempts to investigate the costimulatory properties of thymic epithelial cells have been based on the ability of monolayer cultures of either epithelial cell lines or thymic nursecell preparations to present antigen to mature T cells (Kyewski et al., 1984;Marrack et al., 1989;Lorenz and Allen, 1989). However, none of these preparations has been shown to mediate positive selection and in some cases they actually mediate negative selection (Iwabuchi et al., 1992;Pircher et al., 1993), suggesting either that they are not representative of cells mediating positive selection in vivo or that they have lost the capacity to deliver the signals required for this process. This latter possibility is reinforced by the change in surface phenotype seen when epithelial cells are gro,wn in monolayer cultures. (Nonoyama et al., 1989; our unpublished observations).
To investigate the costimulatory function of epithelial cells under conditions where they can mediate positive selection, we examined their ability to stimulate the proliferation of mature (peripheral) V,88 T cells when presenting the superantigen SEB in reaggregate cultures. Proliferation in response to SEB, like that to conventional antigens, has been shown to require costimulation (Jones et al., 1993). As shown in Fig. 2, no evidence of proliferation, either in terms of increased cell number or of entry of cells into the G2/M phases of the cell cycle, was observed when SEB was presented on thymic epithelial cells in reaggregate cultures. In contrast, marked proliferation in response to SEB was observed in reaggregates of V,88 cells and thymic dendritic cells. This difference was not due to the inability of epithelial cells to bind SEB (Jenkinson et al., 1992) Jenkins and Johnson, 1993). To determine whether the failure of epithelial cells to provide costimulatory signals was associated with the absence of B7, we examined purified epithelial cells for both B7 mRNA and B7 surface expression. B7 was not detectable in thymic epithelial cells by either technique but was expressed by thymic dendritic cells (Fig. 3) correlating with the differing ability of these two cell types to provide functional costimulation. These findings confirm those recently reported by Jones et al. (1993). In addition, because B7-mediated costimulation has been shown to be a requirement for allograft rejection (Turka et al., 1992), they are consistent with our previous observation that grafts of MHC Class II / thymic epithelial cells are capable of prolonged survival in immuno-competent histoincompatible recipients (Ready et al., 1984). Overall, our results suggest that costimulatory signals of the type involved in the activation of mature T cells are not involved in positive selection. However, because TCR signaling alone is insufficient to trigger the maturation of CD4 + 8 thymocytes (Swat et al., 1993), it seems likely that other types of costimulatory or additional signals are involved. Our current studies, using reaggregate cultures of CD4/8 thymocytes and a variety of epithelial and nonepithelial stromal components, suggest that the provision of such signals is uniquely associated with thymic epithelial cells (Anderson et al., in preparation). Identifying these signals should provide an important insight into the control mechanisms for positive selection.  (Jenkinson et al., 1992;Anderson et al., 1993). Briefly, isolated lobes from day 14 fetuses were cultured for 6 days in medium containing dGuo to eliminate lymphoid and dendritic cells, disaggregated and depleted of any residual cells of hemopoietic origin using magnetic beads (Dynal) coated with anti-CD45 (clone M1-9, ATCC). Further depletions were carried out to remove cells express-ing the medullary marker A2B5 before MHC Class II / cells were positively selected on beads coated with anti-Ia d (clone MK-D6, Becton Dickinson).

MATERIALS AND METHODS
Beaded cells were collected and washed four times on a magnet to remove unbound cells and the beads removed enzymically by resuspension in 200/1 of ice-cold Pronase (10 mg/ml, Sigma), followed by a 1-2 min incubation at 37C. After addition of 1 ml of medium containing 10% FCS, liberated cells and beads were separated on a magnet and the cells washed and collected by centrifugation and subjected to a further round of depletion with A2B5 coated beads. Cells isolated in this way express high levels of MHC Class II together with the cortical epithelium markers ERTR-4 and 4F11E and are free of cells expressing macrophage and fibroblast markers .

Preparation of Thymic Dendritic Cells
Thymic dendritic cells were generated from precursors obtained from fetal thymus lobes using a modification of a method described by Inaba et al. (1992) for adult mouse blood. Day 14 fetal thymus lobes were organ cultured for 7 days, teased apart, and the resultant cell suspensions placed in 24-well plates at I x 106 cells/well in medium containing 40 U/ml of recombinant murine GM-CSF (Genzyme). Colonies growing on a monolayer of adherent cells were evident after 1 week and were harvested after 2 to 3 weeks by gentle pipetting. Cells harvested in this way were greater than 90% positive for MHC Class II and expressed the dendritic cell markers NLDC-145 and MIDC-8 (Breel et al., 1987).
Preparation of Reaggregate Cultures TCR-, CD4 + 8 +/CD4-8 + cells were prepared from suspensions of newborn thymocytes by depletion of CD3 cells followed by positive selection on CD8 using antibody coated magnetic beads as described in detail elsewhere (Jenkinson et al., 1992;Anderson et al., 1993). These cells were mixed with purified MHC Class II epithelial cells at a ratio of 2-3:1 and placed as standing drops on nucleopore filters supported on sponge rafts to allow reaggregation into intact thymus lobes. Intact lobes formed within 12 hr and phenotypic maturation, as shown by the appearance of single positive CD4 or CD8 cells with high levels of TCR expression (that is, positively selected cells), was determined after 4 days by three-color flow cytometry on cell suspensions liberated by teasing the lobes apart. To determine the ability of purified epithelial cells to provide costimulatory signals under conditions where they are known to support the positive selection of developing thymocytes, epithelial cell suspensions were mixed with purified V]/8 lymph node T cells at a ratio of 2:1 and allowed to reform reaggregate lobes on nucleopore filters, as described in the previous section. SEB (Sigma) at 10 mg/ml was added to experimental cultures at the outset or omitted in the case of control cultures. For comparison, V]/8 cells were also reaggregated with MHC Class II / thymic dendritic cells in either the presence or absence of SEB. Cultures were harvested after 4 days, mechanically disrupted with fine knives, and counted to determine the total yield of V88 cells.
As a further measure of proliferation, the cell cycle status of T lymphocytes recovered from reaggregate cultures was assessed using the DNA binding dye 7AAD in conjunction with surface labeling for CD4 and CD8 (Rabinovitch et al., 1986).
Cell suspensions were incubated in a mixture of CD4-PE and CD8-FITC (Becton Dickinson), fixed in 80% ethanol, and resuspended in PBS containing 0.1% Tween 20, 0.1 mM EDTA, and 25/g/ml 7AAD (Sigma) 30 min .before analysis. Three-color flow cytometry was carried out on an Elite Dual Laser machine (Coulter) to determine the cycling status of cells expressing either CD4 or CD8. Non-Detection of B7 mRNA by Semiquantitative PCR Expression of B7 mRNA in either purified thymic MHC Class II + epithelial cells or purified (MHC Class II selected) thymic dendritic cells was determined using semiquantitative PCR performed as described in detail elsewhere . In short, cytosolic RNA was extracted from 1-3 x 105 cells, treated with DNase I (Pharmacia), and reversed transcribed using M-MLV reverse transcriptase (Gibco-BRL). B7 and ]%actin sequences were amplified using one cycle at 94C for 5 min and 17-50 cycles at 94C for 30sec, 50C for 60 sec, and 72C for 60 sec. Ten microliters of reaction mix was removed at regular intervals during PCR to encompass the exponential phase of the amplification. PCR fragments, visualized by agarose gel electrophoresis and ethidium bromide staining, were positively identified by size and/or partial DNA sequencing. The intensity of the ethidium bromide stained bands was determined using a gel documentation system (Image Store 5000, UVP, Cambridge, GB) followed by scanning densitometry (E.A.S.Y, UVP). Primer sequences for -actin have been published before  and the primer sequences for B7 correspond to bases 474-492 and 1084-1102 of the published murine B7 cDNA sequence (Freeman et al., 1991). The ,8-actin was used as an internal control for both reverse transcription and the PCR. A negative control incorporating all reagents except cDNA was included in every batch of PCR.
Detection of B7 Surface Expression by Immunolabeling Cell-surface expression of B7 on MHC Class II thymic epithelial or thymic dendritic cells was determined by dual labeling, using a mixture of anti-Iad (Clone MK-D6, Becton Dickinson) and a B7 binding fusion protein between murine CTLA-4 and human Ig (a kind gift from Dr. P. Lane) followed by anti-human Ig-Biotin (Binding Site, Birmingham, U.K.) and a mixture of anti-mouse Ig-FITC (Caltag) and streptavidin-APC (Becton Dickinson). Analysis was by two-color flow cytometry with side and forward scatter gates set to exclude nonviable cells.  (Received July 21, 1993) (Accepted September 23,1993)