Human Thymic Epithelial Cells in Serum-Free Culture: Nature and Effects on Thymocyte Cell Lines

Thymic epithelial cells (TEC) have been cultured for several months and/or for 4 to 5 transfers in a growth factor-defined serum-free medium without concurrent growth of other cell types. The use of monoclonal antibodies and αMAM-6 indicated that the majority of TEC were of medullary origin. The vast majority of cells were positive for LFA-3 and class I, and class II expression, was low or absent. Supernatants from the cultures were shown to contain IL-1ß, IL-6, and M-CSF. Coculture of cloned subpopulations of thymocytes and TEC showed effects of TEC and of secreted ILs on thymocyte proliferation. High percentages of TEC were able to bind DN, DP, or SP thymocyte populations, partly via CD2-LFA-3 adhesion. Thus, it is possible to culture TEC without unknown serum factors and with maintenance of functional activities.


.g., Blackman
et al., 1990;Ramsdell and Fowlkes, 1990; Sprent et al., 1990;von Boehmer and Kisielo, 1990).Further, stromal cells are in a key position for positive and nega- tive selection of developing T-cell precursors.Thus, the thymic epithelial cells (TEC) of the cor- tex seem to select positively via MHC-TCR- peptide interaction, and negative selection is ascribed to bone marrow-derived macro- phages/dendritic cells and tolerance induction to effects of medullary TEC on more mature single- positive (SP) thymocytes (reviewed by Blackman et al., 1990;Boyd and Hugo, 1991).In addition, a variety of interleukins (IL), likely to be of import- ance to T-cell precursor maturation, is known to be secreted by TEC, as reviewed by Haynes   (1990).

To understand the role of TEC in the development of the T-cell repertoire, the nature of the *Corresponding author.Present address: Laboratory for Cellul r Immunology, Department of Medical Anatomy, University of Copenhagen, The Panum Institute, Blegdamsvej 3, DK-2200 Copenhagen N, Denmark.direct interaction between T-cell precursors and TEC, and the importance of secreted factors (hormones/peptides), it seems essential to study well-defined cell populations in culture.Several methods have been used to culture TEC from humans (Sun et al., 1984;Berrih et al., 1985;Blue et al., 1985;Singer et al., 1985;Schuurman et al., 1986;Mizutani et al., 1987;Singer and Haynes, 1987;Galy et al., 1989).In all these reports, fetal calf serum (FCS) or human serum has been used as a necessary constituent of the culture medium.However, it is well known that the addition of serum to the medium has several disadvantages (see also Barnes and Sato, 1980), such as addition of unknown factors (hormones, antibodies, endo- toxins, viruses, and other macromolecules), which may vary from batch to batch.This makes it difficult to isolate and interpret the significance of biological activities executed by the cells and by molecules released by the cells into the culture medium.The culture of TE cells in serum-con- taining medium is further hampered by the ability of the medium to support the growth of macrophages and fibroblasts, the former being critical in an evaluation of T-cell selection, the latter often rapidly overgrowing the epithelial cells.Although several of the previously cited papers advise methods to obtain exclusive growth of TEC, we have not succeeded in making 112 C. ROPKE AND J. ELBROEND cultures in serum-containing medium that were solely constituted by TE cells (unpublished results; Ropke et al., 1990).In the present paper, we have investigated the functional and phenotypical nature of human TEC grown in serum- free culture in the presence of well-defined growth factors.The results indicate that the cul- tured TEC are (1) predominantly of the medul- lary type, (2) bind cells of various thymocyte sub- sets via LFA-3/CD2 adhesion and affect their proliferation, and (3) secrete IL-1, IL-6, and M-CSF.


RESULTS

Characterization of TEC in Serum-Free Culture No significant differences were detected between primary, secondary, terti

y, or qu
rternary cul- tures at a wide range of cell densities at setup as measured by 3H-TdR incorporation and scintil- lation countings, whereas high cell densities were required for establishment of cultures at the fifth transfer.These latter cultures usually showed a low proliferation rate (not shown).By the use of an antibody against cytokeratins 6 and 18 (K 717), the usual finding was that from 95% to 100% of the cultured cells were positive (Fig. la).Negative cells were usually freely growing cells and were most numerous at the start of primary cultures or in the first days after restart of frozen cultures.Another antibody against kera- tins, F 3006, showed similar results, although the staining generally was weaker (not shown).An antibody against cortical TEC (MAS 251) usually showed sparse reactivity in primary cultures.In later cultures, a higher reactivity, both in inten- sity and in numbers of positive cells, was.. present.However, the numbers of positive cells remained in all cultures below 20-30%, and another antibody against cortical epithelium (HB 214), showed no significant reaction in any cul- ture (not shown).The numbers of cells positive for an antibody against medullary TEC (MAS 252) were also low at the start of primary cultures or after rethawing, but in contrast to the previous the numbers of positive cells increased with time and with transfer, and usually 80-95% of the cells were positive in "late" cultures (Fig. lb).These findings indicate that the majority of the cultured cells were of medullary originmor that cyto- keratins characteristic for medullary TEC were increasingly expressed by prolonged culture.

Findings in support for this were obtained by the use of an antibody against outer cells of Hassall's bodies (MAS 256).Although few cells were positive at the start of culture, nearly all cells were positive at the later time intervals, especially in dense cultures (not shown).Also, small accumulation of cells, which showed the features of Hassall's bodies at light microscopy of sections of the cultures (not shown), were always positive.Further, an antibody  against sialomucins (Hilkens et al., 1989), which has shown reactivity with secretory epithelia and Hassall's bodies (Hilkens et al., 1984), showed positive reaction on a high number of cultured cells in most cultures irrespective of their transfer number (Figs. lc and 2).By the use of an anti- body against mesenchymal cells (MAS 253), only very few positive cells were observedmbelow 5% These cells were always found as singles in the area between developing islands of epithelial cells and disappeared with time (not shown).

High percentages of cultured cells were posi- tive for class I and LFA-3 antibodies, whereas the class II expression generally was low or absent.Less than 10% of the cells showed a significant reaction (Fig. 2).

Culture of Thymocyte Cell Lines in the Presence of TEC Frozen thymocytes from the TEC donors were thawed and set up in Terasaki plates a described in Materials and Methods.Maintenance of the cell lines was dependent on the presence of IL-2, which increased the TdR incorporation 25x after 3 days of culture.PHA further increased the incorporation 2x, and Con A depressed the incor- poration of TdR in IL-2 containing cultures about 2x.

To investigate the effects of TEC on the pro- liferation of cloned thymocytes, the following experiments were set up partly to clarify e fects on the IL-2-driven proliferation, and partly to detect a possible comitogenic effect of TEC on PHA-induced proliferation, the latter used as an indicator for functional capability of the cloned thymocytes.

Figure 3 shows the results of a typical experiment, in which variable numbers of TEC were cultured for 3 days together with a thymocyte cell line.As can be seen, a significant depression in TdR incorporation is found by a TEC percent- age at and above 5 in the cultures.

Table 1 shows the results from 21 experiments, in which 19 different cell lines, derived from 3 different donors, were cultured with or without TEC from 5 different donors.As can be seen from the table, a much higher proliferation was found in cultures of thymocytes without admixed TEC than in cultures containing TEC.All ratios were below 1, the average being 0.49+0.18(SD).

In some experiments, the 3H-TdR incorporation in the cultures was measured on each of the 3 days of culture.No significant change in the ratio was found (not shown).

The table shows further that a very significant FIGURE 1. Photomicrographs of cultures of serum-free growth human thymus epithelial cells.Cells are stained by the PAP method after incubation with antibodies gainst (a) cytokeratins 6 and 18 (x126), (b) medullary epithelial cells (x250), and (c) surface antigens of secretory epithelial (x250).

change was found when PHA was added to the cultures.Here, a high 3H-TdR incorporation was found in TEC-containing cultures as compared to cultures without TEC.Thus, the ratio between the cultures was inversed when PHA was added to the cultures, the average of PHA containing cultures being 1.89+0.73 of 13 experiments.

The ratio changed less than 10% in experiments in which cloned thymocytes were grown with either irradiated (2500 rad) or nonirradiated TEC, which i dicates that the previous effects were not dependent on initiation of synthesis of new secreted and/or surface-bound molecules (not shown).

Although it seems likely that the comitogenic effect of TEC on cloned t ymocyte proliferation is due to a directly TEC induced stimulation, parallel to PHA stimulation, an indirect effect mediated via PHA stimulation of TEC--or PHAinduced increased aggregation of cells--cannot be excluded.

The CD antigen composition of the cell lines is given in Table I together with the individual cpm ratios.At the time of the experiments, the stated FIGURE 2. FACS histograms of serum-free grown human thymus epithelial cells (TEC) from secondary cultures.Cells were incubated with a-MHC class I, a-MHC class II, a-LFA-3, or MAM-6 (an antibody against surface antigens of secretory epithelia) antibodies, followed by incubation with FITC-labeled-GAM antiserum.Ordinate: relative cell numbers; abscissa: log fluorescence.Curves on the left-hand side show fluorescence of cells labeled with FITC-GAM only.

CD antigen composition of the lines had remained stable for at least 3 weeks, and no dif- ferences were observed between cells grown with or without TEC, or between cell lines before and after culture (not shown).

Interleukin Secretion from Cultured TEC Supernatants obtained from serum-free cultures of TEC from four donors were tested in an ELISA assay including anti-IL-lo and -IL-lfl.In three of the .supernatantsIL-lfl was detectedn2300, 5800, and 5500 pg/mL, respectivelynand tests for IL- l o were negative (Ropke and Bendtzen, unpub- lished results).To test the importance of IL-lfl for the proliferation of thymocytes+TEC in the pres- ence of PHA, IL-lfl was added to the cultures in variable amounts (Fig. 4).The figure shows that both the proliferation of thymocytes cultured alone and of thymocyte-TEC cultures was increased by addition of IL-1.The proliferation of TEC was not altered significantly by addition of IL-1 (not shown), in opposition to results

reported by Galy et al. (1989).Further, the effects of preculture of TEC in the presence of IL-1 were tested and compared to the effect of direct addition of IL-1 to the cultures with or without bThymocyte lines derived from limiting dilution culture of thymo.cytes.

CIdentity of serum-free grown thymic epithelial cells (TEC), of which 5000 added to each culture of 50,000 cloned thymocytes, grown alone.

dThe ratio between average cpm of cloned thymocytes cultured for days+PHA with TEC without TEC:cpm (thymocytes+TEC)-cpm TEC/cpm thymocytes.

ePercentage CD4/CD8 cells of the various cell lines.grown in 3 days in the presence of IL-2 and the given numbers of thymic epithelial cells (TEC.)Regults are given as percentage values+SD.The mean cpm of cultures without added TEC was set to 100.Cpm's of parallel'cultures of TEC alone are subtracted from the cpm's of cultures of CD4 cells+TEC.Effects o rIL-l-fl on proliferation in 3-day cultures of 50,000 cloned CD4 thymocytes in the presence of PHA.CD4 cells were either cultured alone (thy alone) or in the presence of 5000 thymic epithelial cells (thy+tec).Results are given as mean 3H-TdR incorporation+SD.Cpm's of parallel cultures of 5000 epithelial cells are subtracted from the results of (thy+tec) cultures.addition or rIL-2.Figure 5 shows that, in the absence of IL-2, very high counts are obtained by PHA addition to IL-1 preincubated cultures com- pared with counts from other cultures, and proliferation was negligible in cultures without PHA.The high proliferation rate in cultures including IL-1 preincubated TEC, as compared to other cultures, was maintained after addition of IL-2 to PHA containing cultures, whereas--in the absence of PHA--the prolifer

ion was dim
n- ished in TEC-containing cultures, also with respect to cultures of thymocytes alone, in agree- ment with the previous findings (Table 1).To test if IL-1 was working via IL-6 on thymocyte pro- liferation, experiments, in which oIL-6 was included in some of the cultures, were performed.The results given in Figure 6 show that the increased thymocyte proliferation induced by TEC was abolished by the addition of oIL-6 to the cultures, both with regard to "normal" TEC and TEC preincubated with IL-1.Thus, these lat- ter experiments indicate that the beneficial effect of TEC on proliferation of cloned thymocyte mainly is mediated via secretion of IL-6, a secretion that is augmented by  Experiments, in which otLFA-3 was included in PHA-contain- ing cultures of thymocytes+TEC, showed that the beneficial effects of TEC on thymocyte prolifer- ation was diminished by the addition of this anti- body (not shown), indicating that CD2/LFA-3 adherence (as depicted by "Rosette Formation"; see what follows) is essential for the increase in thymocyte response to PHA induced by TEC. .Effects of preculture for 3 days of thymic epithelial cells (tec) in the presence of rIL-1-]/before coculture with cloned CD4 thymocytes as compared with cocultures of CD4 cells and not-precultured tec.Further, the effects of PHA and IL-2 on cocultures are shown.CD4 cells were either cultured without (A) or with (B) IL-2.In addition, cultures were set up both in the presence (blank columns) and absence (shaded columns) of PHA.Five types of cultures are compared: 1, CD4 cel.ls+precultured tec; 2, CD4 cells+tec+10 p IL-1; 3, CD4 cells+tec; 4, CD4 cells+10 p IL-1; 5, CD4 cells.Results are given as mean cpm+SD of triplicate cultures. Cpm's of parallel cultures of tec are subtracted from the results of cultures of CD4 Histogram showing that anti-IL-6 antibody  abolishes the beneficial effects of thymic epithelial cells (tec) on the PHA-induced proliferation of cloned CD4 thymocytes.Fifty thousand cloned CD4 thymocytes (thy) were cultured for 3 days with or without 5000 thymic epithelial cells in the presence of rIL-2.Tec were either precultured for 3 days in the presence of 10 p rIL-1]/(ILtec) or used directly.To some of the cultures, anti-IL-6 antibody (aIL-6) was added at the start of the cultures.Results are given as mean cpm+SD of triplicate cultures.Cpm's of parallel cultures of tec are subtracted from the results of cultures of thy+tec.

Secretion of M-CSF from Cultured TEC Addition of TEC supernatant to MC cultures of murine bone marrow cells resulted in a number of developing clones after 6 days, a number that was usually between 40% to 70% of that given by the same percentage of supernatant from Con-A stimulated spleen cells (not shown).High num- bers of colonies developed in the MC after addition of cultured TEC directly to the BM cell cultures.Figure 7 shows a typical experiment, in which increasing numbers of TEC were added, and the figure shows that the optimal stimulation of colony formation was obtained after addition of about 20,000 TEC to 100,000 BM cells.Addition of IL-I/to the cultures about doubled the num- bers of colonies induced by TEC (not shown).

Cell morphology after orcein staining of colonies indicated that only macrophage colonies devel- oped in the cultures, pointing out that only M-CSF was secreted by the TEC.microscopy.Table 2 shows results from such experiments in which TEC from seven donors and 24 cell lines from five donors were set up.As can be seen from the table, no significant differ- ences were found between different CD sub- populations in percent rosette formation, all showing a rather high percentage at 4C, whereas the two subpopulations tested at 37 C showed lower percentages.The effects of addition f antibodies against surface molecules to the cultures were evaluated by the use of CD4 / cell line.As can be seen from the table, inter- action between CD2 and LFA-3 seems to be essential for binding of thymocytes to TEC at 4 C.No significant differences in rosette forma- tion were found when a cell line was incubated with TEC from different donors including TEC from the donor of the cell line (not shown).


DISCUSSION

Rosette Formation between TEC and Thymocytes When thymocytes are incubated with TEC, a number of TEC will bind thymocytes in rather stable aggregates--rosettesmboth at 4 C and 37 C.These ros ttes can be counted by light developing in mL methylcellulo'se cultures of 100,000 bone marrow cells after addition of variable numbers of serum-free grown thymus epithelial cells (TEC).Means+SD of triplicate cultures are given.

By the use of the present culture method, it is possible to grow TEC without any appreciable growth of mesenchymal cells, and without any additional separation procedures, as judged from the results obtained with antibodies against cytokeratins.These results indicate further that the majority of cultured TEC belong to a population of secretory cells derived from the thymic med- ulla, although some TEC with cortical phenotype were regularly found.Results from culture of "Thymus epithelial cells (TEC) incubated with cloned thymocytes of the given phenotypes for 1/2 h in the ratio to 10. bAntibodies added to cultures.

Cells incubated at the given temperatures.

Percentage of TEC with bound thymocytes+SD.

'Percentage of thymocyte binding TEC in incubations without added antibodies set to 0.

human TEC in a serum-containing medium point to a mixed cort

al/medullary origin of
the cul- tured TEC (Schuurman et al., 1986;Singer and Haynes, 1987), but cultured cells of subcapsular origin have also been reported (Mizutani et al., 1987).Because of the separate functions in the selection of T-cell clones assigned to cortical and medullary TEC, respectively, as surveyed by Ramsdell and Fowlkes (1990), Sprent et al. (1990), von Boehmer and Kisielow (1990), and Boyd a.nd Hugo (1991), it is of major importance in the evaluation of the T-cell progenitormTEC inter- actions in vitro to clarify the nature of the cul- tured TEC.The present results, indicating predominance of medullary TEC, may show that the culture medium is primarily supportive for med- ullary TEC.However, to our knowledge, no information on possible phenotypic shifts of TEC brought from in vivo to in vitro conditions is reported.Further, in spite of the generally

ccepted view that medullary a
d subcapsular TEC are of ectodermal origin and cortical TEC of endodermal origin (Haynes, 1990), there is no conclusive evidence for this view.Thus, it is possible that cells with potentials for differen- tiation into both cortical and medullary phenotypes/functions are present in the cultures and may differentiate in a given direction depending upon the culture conditions (e.g., presence of ILs or thymocytes).

The serum-free cultured TEC of the present study show significant amounts of class

and LFA-3 m
lecules, but only a limited expression of class II molecules, in accordance with findings in serum-containing cultures, although addition of interferon-' or thymocytes to the cultures aug- ments the amount of class II molecules on TEC (Berrih et al., 1985;Denning et al., 1987).This has not been tested in the present study, but we have shown, by the use of several cell lines derived from different populations of thymocytes, that the cultured TEC affect the IL-2-dependent pro- liferation of these populations, resulting in a depression in 3H-TdR incorporation in the absence of mitogen, whereas TEC augmented the PHA response of the cell lines.This reverse effect seems to indicate that several receptors are involved in the interaction between the cells.Thus, both ICAM-1 and LFA-1 interaction as well as LFA-3 and CD2 interaction are involved in thymocyte-TEC adhesion as shown by rosette formation (Vollger et al., 1987;Nonoyama et al., 1989;Singer et al., 1990; present study).The adhesion is likely to be an essential step for further interactions

and has previously been shown to be
characteristic for early thymocytes expressing rather low amounts of CD3 (Nonoyama et al., 1989).However, in the present study, we show that high percentages of TEC bind both early and late thymocyte subsets, the latter showing CD3 expression comparable to peripheral T cells (not shown).The difference between these studies may be due to that we have tested actively proliferating cell lines rather than G0dl cells, our results being in accordance with those of Singer et al. (1990), who tested Con A-stimulated cells.

Several studies performed with serum-contain- ing cultures have shown that cultured TEC secrete a number of interleukins" IL-1, IL-6, TNF, LIF, and GM-CSF (Le et al., 1987;Galy et al., 1990;Le et al., 1990) of possible importance to the development and differentiation of T-cell precursors.In the present serum-free cultures, we have so far obtained evidence for-secretion of some of these ILs and of effects on T-cell lines of these ILs.Our results indicate that at least a part of the IL-1]/ effect is mediated through stimulation of TEC to secrete IL-6, which acts on the thymocyte subsets as demonstrated by Galy et al. (1990).

Further studies of serum-free cultures are needed to elucidate the constitutive secretion of cultured TEC.This applies also to the present demon- stration of secretion by TEC of M-CSF but not GM-CSF, the latter being demonstrated in serum- containing cultures (Galy et al., 1990;Le et al., 1990).Low amounts of GM-CSF may result in formation of macrophage colonies only (Metcalf, 1977), but the presence of rather high numbers of colonies (Fig. 7) and high numbers of cells in the individual colonies (unpublished results) do n t indicate that this is the reason for the lack of demonstrable GM-CSF.

In conclusion, the present results indicate that the serum-free culture of TEC allows for study of the T-cell precursor-TEC interactions under well- defined conditions as well as a dissection of the pedigree of IL effects in T-cell development.However, we have not in the present study--which was focused on cultures consisting solely of TECmbeen able to test whether the pre- sented effects on thymocyte cell-line proliferation are specific for TEC.This is now being tested in our laboratory.


MATERIALS AND METHODS

Cloning of Thymocytes Material Thymic tissue was obtained from children, ages 1/2 to 2 years, undergoing cardiovascular sur- gery for congenital heart disease.The tissue was cut into approximately 1-2-'mm fragments by the use of surgical knives, and fragments were washed several times in PBS.The washing medium, including the free cells, was collected and the f ee cells were frozen in liquid nitrogen after addition of 10% DMSO to the medium for later establishment of thym

yte cell lines.The tissue fragments were incu
ated for 1 h at 37 C in medium containing 5.0 mg/mL DNAase (Sigma, St. Louis, MO) and 1.5mg/mL collagenase/dispase (Boehringer, Mannheim, FRG) in a culture dish, which was shaken regularly.Thereafter, the sizes of the fragments were considerably below 1 mm3.The fragments were washed two times wi