by Harwood Academic Publishers GmbH Printed in Singapore Mechanisms Involved in the Binding of Thymocytes to

The effects of monoclonal antibodies (mAbs) to cell-surface molecules, divalent cations, and various cell-signaling and metabolic inhibitors on the binding of thymocytes to rat thymic dendritic cells (TDC) were studied using a rosette assay. It was found that TDC/ thymocyte adhesion was stronger and faster at 37C than at 4C. Flow cytometric analysis demonstrated that bound thymocytes were predominantly CD4/CD8 and CD4+CD8-, but in comparison to the phenotype of whole thymocytes, they were enriched in the mature


mposed of non
lymphoid cells and thymocytes (Kyewski et al.,  1982).TDC are also rosette-forming cells (Kyewski  et al., 1982; Landry et al., 1990; Shortman and Vremec,  1991).It is postulated that these close cell-cell con- tacts are responsible for most of TDC functions (Adkins et al., 1986;  Kyewski, 1988).Although in- teractions between peripheral dendritic cells (DC) and T lymphocytes have been extensively explored, little is known about the mechanisms involved in TDC/thymocyte binding.This question is addressed in the present work using an original system for isolation of rat TDC.


RESULTS


Dynamics of Thymocyte Adhesion to Rat TDC

To study adhesion characteristics of rat TDC, we first used an alternative method for their isolation and purification (Ilid et l., in press).This inc uded isola- tion of low-density cells from thymocyte suspension over a Nycodenz gradient (density 1.078 g/cm3; osmolarity 390 mOsm) and their subsequent culti- vation with 20% TE-R 2.5 + HT supernatant obtained by cocultivation of a medullary epithelial cell line (TE-R 2.5) (olid et al., 1992) and hydrocortisone- resistant thymocytes.This procedure allowed the re- covery of TDC of relative high purity (more than 80%), their differentiation and expression of most markers characteristic for rat TDC in situ, such as MHC class I and class II molecules, CD45, CD25, LFA-1 (CD11a/CD18), ICAM-1 (CD54), and OX-44   (CD53).Certain TDC subsets expressed CD11b and th

ocyte ma
kers Thyl, CD2, CD4, and CD8.These cells were functionally very active in inducing strong proliferation of autologous thymocytes depleted of endogenous accessory cells even without any addi- tional stimuli (Ilid et al., in press).

Such prepared TDC were incubated with syn- Dynamics of rosette formation between TDC and thymocytes.TDC were mixed with thymocytes (ratio 1:20) and incubated in Terasaki plates (hanging drop assay) for different periods of time both at 37 or 4C.The percentages of TDC-forming rosettes from one of three similar experiments are given.

geneic AO rat thymocytes (ratio 1:20) both at 37 and 4C using a hanging-drop assay in Terasaki plates.

After different periods of time, the percentage of TDC-forming rosettes were calculated.The results presented in Fig. 1 show that most TDC-formed rosettes with thymocytes at 37C.The binding was very strong because mostly large cell clusters were observed (Fig. 2).The adhesion process was also very fast reaching the maximum as early as after 30 min of cell incubation and slightly decreased thereafter (up to 5 hr).In contrast, the initial binding (30 min)  at 4C was very low.After that, it progressively in- creased (up to 3 hr) reaching almost the same values as those observed at 37C (Fig. 1).


Phenotypic Characteristics of Thymocytes

Bound to TDC

We next tested the phenotype of thymocytes bound to TDC by flow cytometry.Purification of rosettes, detachment of thymocytes from TDC, and staining procedure are described in Materials and Methods.Thymocytes were easily identified by appropri- ate g

ing.The results presented in Fig. 3 and T
ble 1 show that TDC predominantly clustered with CD4/CD8 and CD4/CD8 thymocytes.However, the percentage of CD4/CD8 cells was lower, whereas FIGURE 2. TDC rosettes stained with hematoxylin-eosin.TDC rosettes were formed by incubating TDC and thymocytes (ratio   1:20) for 30 min at 37C as described in Materials and Methods.Before staining rosettes were purified by brief centrifugation (50 g for min at 4C) over FCS and incubated for 20 min on PLL-coated glass slides to allow their better attachment and spreading, followed by 0.1% glutar-aldehide fixation.Magnifi- cation x320.


A B

CD4 fluorescence FIGURE 3. Flow cytometric an lysis of thymocyte subsets defined by the expression of CD4 and CD8 among total thymocytes (A)  and thymocytes bound to TDC in vitro (B).The formation of rosettes, detachment of thymocytes from TDC, and staining of thymocytes with mAbs were as described in Materials and Methods.Thymocytes were analyzed on a FACScan flow cytometry by appropriate gating.Representative profiles of CD4 and CD8 expression are given.and thymocytes bound to TDC in vitro (dotted line).Thymocytes were prepared and stained by R 73 mAb as described.The his- tograms are representative of three experiments with similar re- sults.The vertical bar represents the level of nonspecific stain- ing using PBS and secondary antibody s control.

the percentage of CD4/CD8 cells was higher than the values of these cells subsets in the whole thymocyte population.The conclusion that thymocytes bound to TD were enriched in more mature thymocytes was drawn by analyzing the fluores- cence profile of thymocytes stained by an anti- c[TCR antibody (R73).Figure 4 and Table 1 show that among adherent thymocytes (compared to whole thymocytes), a higher percentage of the 0TCR hi subset was observed.

Effect of Bivalent Cations on Rosette Formation between TDC and Thymocytes

The next experiments were designed to study the mechanisms involved in TDC/thymocyte adhesion.

We first studied the effect of bivale t cations (Ca 2+   and Mg2/) in the rosette formation after 30 min of cell incubation at 37C. Figure 5 shows that Ca 2+- and Mg2/-free HBSS medium only partly suppressed thymocyte binding to TDC.The separate addition of Ca 2+ and Mg 2/ ions to HBSS medium partly re- stored thymocyte binding to TDC.If these cations were added simultaneously, the percentage of ro

settes did not significantly differ from the values ob
ained using the classical medium (RPMI+10% FCS).

The effect of Ca 2+ and Mg 2+ chelators (EDTA and EGTA) was also tested both at 37 and 4C. Figure 6 shows that these agents in a dose-dependent man- ner (except at 4C, after 3 hr) suppressed rosette for- mation.The percentage of inhibition was almost equal independently of incubation temperature.However, the inhibition decreased with prolonged incubation (3 hr).

Signaling Pathways Involved in TDC/Thymocyte Adhesion Signaling pathways involved in the rosette for- mation between TDC and thymocytes were stu- died using various types of metabolic inhibitors.

Figure 7 shows that cytochalasin B (an inhibitor of microfilament formation) partially decreased the percentage of rosettes at 37C (30 min).Similar re- sults were obtained after 3 hr of cell incubation (not shown).As expected, cyclohex

ide (an inhibitor
f protein synthesis) did not influence the TDC/ thymocyte binding.We also treated TDC and thymocytes with various protein kinases inhibitors at 37C.H7 (a PKA and PKC inhibitor) and genistein (a tyrosine kinase inhibitor) did not influence the adhesion process.In contrast, W7 (specific inhibitor of calmodulin-dependent protein kinase) was partly inhibitory.Similar inhibition was seen after 3 hr (not shown).The adhesion was not modified by using % rosettes FIGURE 5. Effect of bivalent cations on thymocyte binding to TDC.Thymocytes were mixed with TDC in different media.Ca 2+ and Mg were used at concentration of 5 mM.Cells were incubated for 30 min at 37C.The results are presented as the mean percentages of rosettes + SD from 3 different experiments.% rosettes FIGURE 6.Effect of bivalent cation chelators (EGTA + EDTA) on thymocyte binding to TDC.Thymocytes and TDC were incu- bated in RPMI/10% FCS at 4 or 37C for 30 min or 3 hr.EDTA and EGTA were added at concentrat ons of 1-10 mM.The results are presented as mean of rosette percentages + SD from 3 different experiments.Na-orthovanadate 100 120 rosettes FIGURE 7. Effect of various metabolic and cell-signaling inhibitors on thymocyte binding to TDC.TDC and thymocytes were preincubated with different inhibitors for 30 min at 37C before mixing as described.Cells wer then incubated with inhibitors for additional 30 min at 37C.The results are presented as the mean percentages of rosettes + SD from 3 different experiments.p < 0.001 compared to control (RPMI/10% FCS edium).sodium orthovanadate (an inhibitor of protein phosphotyrosine phosphatases).

Surface Cell Molecules Involved in TDC/


Thymocyte Binding

The final aim of this work was to examine the role of cell-surface molecules in TDC/thymocytes adhe- sion by using mAbs directed to several rat adhesion molecules.Figure 8 shows the effect of mAbs to [32 integrins, ICAM-1, and CD2 on TDC/thy

cyte bindin
.The results demonstrated that of the mAbs, WT.1 (anti-LFA-1), WT.3 (anti-CD18), and 1A29 (anti- ICAM-1) mAbs were partly inhibitory after 30 min.WT-3 had the strongest effect (approx.55% inhibi- tion) that was not further potentiated using combi- nation of WT.1 and WT.3. Figure 8 also shows that inhibitory effects of WT.1 and WT.3 were lesser after 3 hr of cell incubation.At that time, point 1A29 was not inhibitory.The mAbs to CD11b (ED7) or CD11b/c (OX-42) as well as CD2 (OX-34) did not significantly modify the adhesion.The inhibitory effect of OX-34 mAb on TDC/thymocyte adhesion was also not seen under condition when the main pathway (LFA-1/CD18) was blocked using a com- bination of WT.1, WT.3, and OX-34 mAbs.A lot of mAbs stimulated TDC/thymocyte bind- ing.They were OX-3 and OX-6 (both directed to class II MHC molecules), R73 (anti-0[3TCR), W3/25 (anti- CD4), OX-8 (anti-CD8), and G3C5 (anti-CD45 frame- work) antibody (Fig. 9).The gtimulated adhesion was not only manifested by an increase in the percent- age of rosettes, but also by a significant enlargement of their sizes.

The stimulatory effect o anti-0[3TCR, anti-CD4, and anti-CD8 was visible during the first adhesion phase (30 min), whereas the effect of OX-3 and OX- 6 mAbs was observed after 3 hr of cell incubation.

G3C5 mAb stimulated TDC/thymocyte binding in both adhesion phases.OX-18 mAb, directed to class I MHC, was without significant effect in this process.We further tested whether stimulatory effects of these mAbs were mediated by LFA-1.The results presented in Fig. 9 show that, except for G3C5, preincubation of thymocytes with WT. 1 abrogated the stimulated adhesion of all these mAbs.


DISCUSSION

This work was designed to study the mechanisms involved in the adhesion between TDC and thymocytes.We started the experiments using a modification of the procedure for isolation of rat TDC.It involved selective centrifugation of whole thymocyte suspension over a Nycodenz gradient (14.5%, osmolarity 390 mOsm) and further cultiva- tion of low-density cells (TDC purity about 30%) for 3 days in medium supplemented with the TE-R 2.5 + HT supernatant.This supernatant was prepared by cocultivation of a rat medullary thymic epithelial cell line (TE-R 2.5) (olid et al., 1992) with hydro- cortisone-resistant thymocytes.Cytokines and other soluble factors derived from these cells promoted significant survival of TDC in culture and their morphological and phenotypic l differentiation to the cells expressing most characteristics of TDC in situ.At the same time, TDC purity significantly increased (more than 80%) due to apoptosis of con- taminating thymocytes and selective attachment of thymic Mq) and monocytelike cells to plastic.All these properties of TDC were presented n our recent work (Ilid et al., in press).

Using such prepared cells, we found that rat thymocytes strongly bound to TDC and formed large clusters (rosettes).Attached thymocytes were en- riched in the 0TCR hi subset (predominantly CD4+CD8-), but a large proportion of thymocytes were phenotypically immature, CD4/CD8 cells.Such an in vitro analysis of thymocyte phenotype has not been perfo med.However, the results are simi- lar to those of Shortman and Vremec (1991) who analyzed by flow cytometry the phenotype of thymocytes bound to mouse TDC, immediately after in vivo isolation of the rosettes.They found that the TDC rosettes were enriched in thymocytes expressing high levels of surface TCR and CD3, and these included both CD4+CD8-CD3 and CD4- CD8/CD3 mature thymocytes.Both these and our results are in agreement with the topographical localization of TDC in situ (cortico-medullary zone and medulla) in which final differentiation and maturation processes of thymocytes occur (Barclay  and Mayrhofer, 1981; Hamblin and Edgeworth,   1988).The finding that CD4+CD8+czTCR w (imma- ture phenotype) thymocytes showed a similar in- cidence in total thymocyte pool as the population isolated from TDC after binding, suggests that the immature thymocytes in the cortex have already de- veloped the potential to bind TDC that they may encounter later on during their intrathyrnic sojourn.

Physiologica

y,
hymocytes with this phenotype can be localized in the cortico-medullary zone where they may come into close contact with TDC.Thymocytes and TDC were prepared as described in Materials and Methods.Before mixing, thymocytes were preincubated for 30 min at 4C with mAbs WT.1, WT.3, OX- 34, combination of these mAbs or BH1 (an irrelevant [Ir] mAb) and TDC were preincubated with 1A29, ED7, OX-42 mAbs or BH1, all at the concentrations of 10 gg/ml.Values (mean + SD from three to four different experiments) are given as percentage relative binding to control (medium without mAb) (100% relative binding).p < 0.01; p < 0.001 compared to Ir mAb.medium IrmAb WT. : : i : i : !: i : i : !: !: i : !: !: i : i : i : i : !: i : : i : i : i : !: i : i : !: i : i : !: i i ] --:'* R73+ WT.1 wa/ s . . .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: .: !: .: .: .: .: .: .: .: .: !: .: .: .: .: .: .: .: .: : .: .: .W3/25 + WT.

',','

OX-8

OX-8 + WT. : : i : i : i : i : i : i : i : i : i : !: i : i : i : !: i : i : i : i : !: !: !: i : !: !: l " G3C5+ WT. % relative binding FIGURE 9. Effect of mAbs reactive with rat cell-surface antigens on thymocyte binding to TDC.Thymocytes and TDC were pre- pared as described in Materials and Methods.Before mixing, thymocytes were preincubated for 30 min at 4C with mAbs R 73, W3 / 25, OX-8, G3C5, WT.1, IrmAbs (BH1 or BH2 mAb), or WT.1 in combination with these mAbs, and TDC were preincubated with OX- 3, OX-6, OX-18 mAbs, or BH1, all at the concentration of 10 tg/ml.Values (mean + SD from three to four different experiments) are given as percentage relative binding to control (medium without mAb) (100% relative binding).p < 0.01; p < 0.001 compared to Ir mAb (BH1).Values for BH2 were almost the same as BH1 (data not shown).contact of TDC with CD4/CD8 is also possible in the medulla.Using double immunofluorescence, we found that 15-20% of thymocytes in the medulla of AO rat thymus were CD4/CD8 (oliG unpublished observation).Some immunohistological investiga- tions also showed preferential contacts of CD4 T ells with DC in situ (Janossy et al., 1980), but more precise dual immunolabeling was not performed to demonstrate whether they are double (CD4/CD8/) or single (CD4+CD8-) positive.We think that TDC have different impact on distinct thymocyte sub- sets (induction of apoptosis of CD4/CD8 cells; stimulation of proliferation of CD4+CD8 cells).This hypothesis is currently being tested in our laboratory.

The main aspect of this work concerned the mechanisms involved in the binding of hymocytes to TDC.Such studies are scarce, but it is well known that DC isolated from peripheral lymphoid tissues efficiently cluster T cells, a phenomenon that is not dependent on antigen-specific interaction, but rather engagement of various adhesion molecules (Inaba  et al., 1989; King and Katz, 1989).Our findings in- dicate that the LFA-1/ICAM-1 interaction plays a key role in the early adhesion phase (30 min) be- tween thymocytes and TDC, when the adhesion process was maximal.Inhibitory effects of mAbs against these molecules decreased after prolonged incubation (3 hr).This is in accordance wi h previous studies in other cells systems showing the role of the LFA-1 molecule in the stabilization of the early rapid adhesion process (Dustin and Springer, 1989;   Lepesant et al., 1990).Scheeren et al. (1991) demon- strated that clustering of human peripheral blood DC with T lymphocytes was predominantly LFA- 1/ICAM-l-dependent.However, in contrast to our results, the inhibitory effect of applied mAbs was seen after 4 hr of cell incubation, at the time point of maximal adhesion.The difference could be ex- plained by different dynamics of cell adhes on in these two systems, differences in T cells used (thymocytes vs. peripheral T cells) and probably different phenotypic and functional status of DC.

In our experiments, his adhesion pathway was probably a consequence of the binding of LFA-1 ex- pressed on thymocytes to ICAM-1 expressed on TDC, but an engagement of LFA-1 on TDC with a subset of thymocytes expressing ICAM-1 could be also considered.Namely, in our previous studies, we found that rat TDC express both LFA-1 and ICAM-1 (Ilid et al., in press), whereas almost all thymocytes and a subset of these cells (20%) are LFA-1 and ICAM-I/, respectively (t2olid et al., 1994).

Our results also demonstrated that anti-LFA-1 and anti-CD18 mAbs had stronger inhibitory effects than anti-ICAM-1 mAb, suggesting that LFA-1/CD18 might use other ligands (ICAM-2 or ICAM-3) (Staunton et al. 1989; Fawcett et al., 1992).However, little is known about their expression on TDC and mAbs to these molecules in rat are not avail- able.The importance of the LFA-1/ICAM-1 pathway in our experiments was also documented by the dependence of cell clustering upon Ca R+ and Mg 2+ ions.It is well known that these cations are neces- sary for 2 integrin function (Patarroyo et al., 1990;   Springer, 1990).

Significance of the LFA-1/ICAM-1 pathway in thymocyte maturation is well documented because mAbs to these molecules block the development of DP thymocytes in fetal thymic organ culture (Fine  and Kruisbeek, 1991) and antigen-dependent de e- tion of DP thymocytes from mice that are trans- genic for class I MHC-restricted TCRs (Carlow  et al., 1992; Pircher et al., 1993).The latter findings are of special interest because TDC are believed to play a crucial role in negative selection (deletion of autoreactive T cells in the thymus) (Ramsdell and   Fowlkes, 1990).

We demonstrated that TDC/thymocyte binding was not completely blocked by anti-LFA-1, anti- CD18, and anti-ICAM-1 mAbs, suggesting that other adhesion molecules are involved.MAbs against CD11b were not inhibitory, which is in contrast with their effect on clustering between peripheral rat DC and T lymphocytes (Damoiseaux,  1991) or thymocytes and thymic macrophages (M)   (olid et al., 1994).One possible explanation of the phenomenon is a low expression of this [2 integrin on TDC (Ili et al., in press).Scheeren et al. (1991)  also reported that adhesion between peripheral DC and T cells was partly LFA-l-independent because T cells from leukocyte adhesion deficiency (LAD), which are defective in ]2 integrin expression, bound to DC at low rates.We did not identify other adhe- sion molecules because numerous mAbs against different surface-cell molecules w re not inhibitory.It was important to stress that OX-34 (anti-CD2) mAb used in this study as well as other anti-CD2 mAbs (OX-54 and OX-55) (data not shown) did not inhibit rosette formation, suggesting that the CD2/CD48 interaction (Killen et al., 1988) was not probably operative under these experimental conditions.It was demonstrated that the CD2/LFA-3 pathway is not involved in T-cell/blood-DC (Scheeren et al.,  1991) binding, which is in contrast to the situation of T-cell adhesion to tonsillar DC, where this path-way has been found to play a role (King and Katz,   1989).

A number of adhesion molecules has been iden- tified on DC such as B7, B7-2, VCAM-1, CD44 (King  and Katz, 1989; Scheeren et al., 1991; Caux t al., 1994;  Inaba et al., 1994), but their role in binding of thymocytes to TDC has not been studied.We found that mAbs against CD4, CD8, {x3TCR, and class II MHC molecules stimulated TDC / thymocyte adhe- sion.The process was probably not a consequence of a simple crosslinking of mAbs on cell surface because the stimulated adhesion was not seen at 4C

(not shown) and was inhibited by LFA-1 (Fig. 9).In addition, an enhancement of clustering was also observed when both TDC and thymocytes were pr incubated with saturated concentrations of the mAbs before cell mixing (data not shown).It could be postulated that under these conditions, both spe- cific receptors and maybe Fc receptors on TDC (if they exist) are occupied precluding the possibility of crosslinking.Many mAbs, with different isotypes (IgG1 or IgM), to rat cell-surface molecules that are abundantly expressed on thymocytes, such as anti- CD2 (Fig. 8) or anti-CD43 and OX-52 (not shown), were not stimulatory.This is another argument favoring the specificity of the observed phenomenon.

The stimulated adhesion could be explained by previous results that showed that signaling through TCR and coreceptor molecule triggered by specific mAbs to these antigens increased affinity of LFA-1 for its ligand (Dustin and Springer, 1989; van Kooyk  et al., 1989).The effect was not a consequence of the upregulation of the LFA-1 expression on cell surface, but rather was a result of conformational changes of the LFA-1 molecule.A similar effect of mAbs to CD3, CD4, and CD8 was observed in another system of heterotypic cell adhesion using thymocytes and thymic epithelial cell lines (Lepesant et al., 1990).Certain mAb to class II MHC induced homotypic aggregation of lymphocytes that was also partly FA-l-dependent (Kansas and Tedder, 1991).

Our preliminary results showed that all these stimulatory antibodies modify intracellular signaling pathways of thymocytes, but their effects on thymocyte proliferation in the presence of TDC are not the same.Namely, R73 stimulated thymocyte pro- liferation, whereas the others were inhibitory.The results are in agreement with those reported by Xu et al. (1992), who did not show the correlation between the intensity of human blood DC/T lymphocyte clustering and proliferation rates of the T cells.

In c ntrast to the previous mAbs, an anti-CD45 framework mAb (G3C5) stimulated the adhesion via an LFA-l-independent pathway.This antibody, which has been recently produced in our laboratory (Pavlovid et al., manuscript in preparation), induces strong homotyipic aggregation of rat leukocytes.It was also found that certain anti-human CD45 mAbs either increased the size of clusters in culture of DC and huma T lymphocytes (Xu et al., 1992) as well as tonsillar T cells and U-937 cells (King et al., 1990)  or reduced cluster stability between DC and T lymphocytes (Prickett and Hart, 1990).G3C5 is dif- ferent from similar proaggregatory anti-CD45 mAbs (Lorenz et al., 1993; Bernard et al., 1994) by its cap- ability to induce the LFA-l-independent adhesion of both resting and activated leukocytes and at the same time to stimulate mitogen-induced T- lymphocyte proliferation and allogeneic MLR.The effect is probab y epitop-specific and is not influ- enced by isotype of the mAb (IgM) (Pavlovid et al.,  manuscript in preparation).The mechanisms in- volved in the processes are currently investigated Scheeren et al. (1991) also demonstrated that an anti- CD44 mAb enhanced conjugate formation between T cells and blood DC that could not be blocked by anti-LFA-1 mA

t an int
ct cytoskeleton and the activity of calmodulin-dependent protein kinase were partly responsible for TDC/thymocyte adhe- si n because the process was blocked by cyto- chalasin B and W7, respectively.This is in agreement with the results published for blood DC/T-cell bind- ing (Scheeren et al., 1991).We hypothesize that W7 inhibits cell adhesion by modulating cytoskeleton integrity, because similar observations were published in our recent study dealing with the LFA-1- dependent homotypic aggregation of rat leukocytes (Pavlovid et al., 1994).

Blood DC/T-cell adhesion was also partly blocked by H7 (an inhibitor of PKA and PKC) (Scheeren   et al., 1991), which was not seen in our experiments.We also found that neither inhibition of protein tyrosine kinases by genistein nor inhibition of protein tyrosine phosphatases by sodium ortho- vanadate blocked rosette formation.These enzyme inhibitors have not been tested in other similar cell systems.

We also found that the kinetics of rosette forma- tion differ between 4 and 37C.Both processes were partly dependent on Ca 2/ and Mg 2+ ions.At the same tim , anti-LFA-1/CD18 mAbs were without significant effect at this temperature (data not shown).At the moment, it is not clear whether this difference is due to difference in binding characteristics and/or MIODRAG OLI and VESNA ILId dependent on lower-cell motility at 4C.So this phenomenon needs to be explored in more detail in further experiments.

In conclusion, this work shows that thymocyte adhesion to TDC is a complex and poorly under- stood process and underlines some differences be- tween our results and those publi

tho
anadate, an inhibitor of phosphotyrosine protein phosphatases; cytochalasin B, an inhibitor of microfilament formation; and cycloheximide, an inhibitor of protein synthesis.All chemicals were obtained from Sigma (USA .Their concentrations used in the experiments were nontoxic for cells, studied by tripan blue dye exclusion.

Isolation of TDC Thymic-cell suspension was obtained by teasing thym ses against a steel mesh.Released cells were collected and resuspended in RPMI-1640 medium (Serva, Munich) containing 10% fetal calf serum (FCS) (Flow, Irving, Scotland), 2 mM 1-glutamine, and 1% gentamycin with addition of 0.04% EDTA to dissociate thymic clusters.The cell suspension was filtered through a nylon gauze to remove fibrous residue.To enrich this suspension for TDC, the cells resuspended in the same medium (2 x 108 cells/3 ml) were layered above 3 ml of Nycodenz gradient (den- sity 1.078 g/cm and osmolarity 390 mOsm) and cen- trifuged at 600 g for 15 min at room temperature.Cells from the interface zone were collected, washed twice, adjusted at 2.5 x 10 cells/ml and cultivated for 3 days in RPMI/10% FCS medium with addition of 20% TE-R 2.5 + HT supernatant (24-well, flat bottom plates; Flow) in an incubator with 5% CO 2. The TE-R 2.5 + HT supernatant was prepared by cocultivating a confluent monolayer of a rat thymic medullary epithelial cell line TE-R 2.5 (olid et al.,  1992 and hydrocortisone-resistant thymocytes as described (Ili4 et al., in press).After this culture period, TDC became nonadherent and formed large clusters.Contaminating McI) became adherent whereas most thymocytes died.Nonadherent cells  (Tamatani and Miyasaka, 1990)   and (Tamatani et al., 1991a)b.D. mAb obtained rom Dr. C. Dijkstra, Amsterdam (Damoiseaux et al., 1989).E. mAb obtained from Dr. T. Hunig (Wurzburg) (Hunig et al., 1989).

were collected, resuspended in RPMI/10% FCS medium with 0.04% EDTA, to dissociate clusters, and purified again over a Nycodenz gradient as de- scribed before.The purity of such prepared TDC was usually more than 80%.

For statistical analysis (Student test), the percentage of relative binding in the presence of specific mAb was compared with that using isotype corre- sponding to irrelevant mAb.

Rosette Assay for TDC-Thymocyte Binding Cultivated TDC (1 x 104) were mixed with 2 x 105 thymocytes (ratio 1:20) in 20 tl of RPMI/10% FCS medium in Terasaki microwell plates and then cul- tivated in hanging drops at 4 or 37C for various periods of time.The cells were observed under a light microscope.TDC that bound 4 or more thymocytes were scored as rosettes.For each assay, 200 TDC were counted and each determination was performed in duplicate.The results are given as the mean percent-