Eur. J. Immunol. 1989.19: 649-653 In vifro differentiation of immature thymocytes 649

Janko NikoliC-ZugiC, Mark W. Moore and Michael J. Bevan

Department of Immunology, Research Institute of Scripps La Jolla

Characterization of the subset of immature thymocytes which can undergo rapid in vitro differentiation*

clinic, Recently, we reported that thymocytes expressing the CD8 molecule on their surface can give rise to CD4'CD8' double-positive and CD4' single-positive progeny follow- ing intrathymic transfer into an irradiated host mouse. Thymcoytes expressing a high density of CD8, referred to as CD8hi, and those expressing a low density of the molecule, CD81°, were both able to differentiate in vivo. In this study we examined the ability of these CD8' thymocyte populations and of CD4-CD8- double-negative thymocytes to change their phenotype during brief in vitro culture. CD8' thymocytes were prepared by anti-CD4 plus complement lysis followed by positive selection of the survivors on anti-CD8-coated plates. After 16 h of culture, >60% of CD8' thymocytes became double-positive. Both CD8hi and CD8'" cells were able to show this in vitro change: about 30% of the former and about 80% of the latter became double-positive. In contrast to this, double-negative thymocytes which had been depleted of cells expressing low densities of CD8 did not show such a phenotypic conversion in vitro. Further panning experiments suggested that all of the CD8" thymocytes actually express a low surface density of the CD4 molecule which is undetectable in our cytofluorometric assays.

1 Introduction

In the thymus mature CD4' and CD8' cells differentiate from CD4-CD8- double-negative (DN) precursors [l] via an ob- scure series of events. In the course of their differentiation T cell precursors undergo various phenotypic changes and are selected to recognize foreign antigen in the context of self major histocompatibility complex (MHC)-encoded molecules. In addition, negative selection mechanisms eliminate T cells which possess dangerously high affinity for self MHC and the self peptides presumed to be associated with them. The entire process results in the emergence of self MHC-restricted, self tolerant T lymphocytes expressing high levels of functional T cell receptor (TcR) and CD4 or CD8 accessory molecules. While it is generally believed that the TcR is the main struc- ture guiding both positive and negative selection, the acces- sory molecules CD4 and CD8 also have a role.

Cross-correlation of various markers on thymocyte subpopula- tions by fluorometric analysis has yielded useful information pertinent to possible developmental lineages in the thymus [2-51. Precise establishment of direct precursor-product rela- tionships has been achieved by several in vivo and in vitro studies. Studies from this laboratory have shown that the heat- stable antigen-positive (HSAg") DN thymocytes contain pre-

[I 72421

* This study was supported by National Institutes of Health grants CA 25803 and A1 07244.

Correspondence: Janko NikoliC-ZugiC, Department of Immunology, IM4, Research Institute of Scripps Clinic, 10666 North Torrey Pines Road, La Jolla, CA 92037, USA

Abbreviations: DN: Double-negative D P Double-positive FACS: Fluorescence-activated cell sorter FCS: Fetal calf serum FITC: Fluorescein isothiocyanate GaRIg: Goat anti-rat Ig HSAg: Heat-stable antigen IL 2R: Interleukin 2 receptor MHC: Major histocompatibility complex PE: Phycoerythrin TcR: T cell receptor

cursors of mature thymocytes [6] and that HSAg' DN cells transiently express IL 2 receptor (IL 2R) prior to undergoing further differentiation [7]. CD8 expression precedes that of CD4 during fetal development [8] and on intrathymic transfer of adult DN cells [7]. Furthermore we have recently shown that purified CD8' cells transferred intrathymically give rise to double-positive (DP) and CD4' single-positive progeny [9]. CD8'CD4' thymocytes express different densities of this molecule, and both CD8hi and CD81° cells (the latter appear DN by FACS analysis but are distinct from true DN thymo- cytes in numerous ways) differentiate in the thymus after intrathymic injection [9]. These findings indicate that function- ally and phenotypically immature subset(s) of CD8' thymo- cytes are on a productive pathway of T cell maturation. In this report we further charactrize immature CD8' thymocytes and their immediate in vitro descendants. We show that these cells already express low levels of the CD4 molecule on their sur- face, and that they yield DP progeny in vitro. This is in con- trast to the true DN thymocytes, which are incapable of a similar in vitro differentiation.

2 Materials and methods

2.1 Mice

BALBkByJ, BALB.B and C57BL/6 animals were bred and maintained at Scripps Clinic and Research Foundation animal facility. Mice were used at 4-8 weeks of age.

2.2 Reagents

The rat monoclonal antibodies (mAb) RL172.4, anti-CD4 [lo], 3.168, anti-CD8 [ l l ] and PC61, anti-IL2R [12] were pro- duced in our laboratory in the form of ascites fluid. The latter two were partially purified by 50% saturated ammonium sul- fate salt precipitation and column chromatography and conju- gated to fluorescein isothiocyanate (FITC) as described earlier [9]. Purified goat anti-rat Ig (GaRIg) was obtained from Sigma Chemical Co. (St. Louis, MO). It was absorbed on mouse Ig

0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1989 0014-2980/89/0404-0649$02.50/0

650 J. NikoliC-ZugiC, M. W. Moore and M. J. Bevan

coupled to Sepharose and conjugated to FITC in our labora- tory. A phycoerythrin (PE) conjugate of rat anti-CD4 mAb, GK1.5 [13] and biotinylated rat anti-CD8, 53.6 [14] were purchased from Becton Dickinson (Sunnyvale, CA). Guinea pig serum was used as a source of complement.

Eur. J. Immunol. 1989.19: 649-653

2.3 Cell preparation

CD8+ thymocytes were prepared as described earlier [9]. Briefly, CD4hi cells were eliminated by two cycles of antibody (RL172.4) and complement treatment, followed by recovery of viable cells (routinely < 4% of the starting population) by centrifugation over FicoWsodium-diatrizoate (Pharmacia, Uppsala, Sweden). CD8' cells were isolated by positive selec- tion on petri dishes coated with 1 : 150 dilution of 3.168 ascites in phosphate-buffered saline (PBS) [15]. Cells (4 x 10') were incubated on coated dishes (Fisher # 08-757-12, Pittsburgh, PA) for 1 h at 4°C. Nonadherent cells were removed by exten- sive gentle washing with Hanks' balanced salt solution/2% fetal calf serum (HBSS/2% FCS). When no unattached cells were observed under the microscope (typically after 12-15 washes), adherent cells were recovered by vigorous pipetting. A similar anning procedure was used to obtain CD4'" cells

coated with GaRIg (1 : 30 in PBS). This treatment selected for the cells which survived anti-CD4 plus complement treatment but were coated with the rat anti-CD4 antibody. For negative selection by panning, only the nonadherent cells were col- lected, without further washing of plates.

from CD4 P -depleted thymocytes. In this case the dishes were

2.4 Cytofluorometric sorting and analysis

For sorting experiments, anti-CD4 plus complement-killed, anti-CD8-panned cells were separated into CDShi and CD8" subsets according to the intensity of fluorecence following staining with FITC-3.168. A FACStar instrument (Becton Dickinson) was used to obtain the indicated percentages of brightest and dimmest cells. Cross-contamination was < 2%. Dead cells were excluded by selective gating. Sorting for CD8" thymocytes was performed to include only the cells which appeared CD8- (as determined by control staining with an irrelevant FITC conjugate). For cytofluorometric analysis, lo6 cells were stained with saturating amounts of directly con- jugated antibodies for 20 min at 4 "C. After extensive washing, cells were resuspended in 1% formaldehyde fixative and ana- lyzed on a FACS IV instrument (Becton Dickinson). Figures represent 1 X 104-5 x lo4 events scored per sample. Dead cells were excluded by selective gating. For both the analysis and sorting, background fluorescence was determined from that obtained from unstained cells or from cells stained with irrelevant fluorochrome-conjugated antibodies.

2.5 In vifro cultivation of thymocytes

Subsets of thymocytes were cultured in RPMI 1640 sup- plemented with 20 mM L-glutamine, 50 pM 2-mercap- toethanol, 5 mM Hepes, antibiotics and 10% FCS (Gemini, Calabasus, CA). Cells were plated at 1 x 106/ml in 24-well plates (Costar #3242 Mk 11, Cambridge, MA), and were kept at 37°C in 7% C02/air. At the indicated times cells were counted, scored for viability by trypan blue exclusion and phenotypically analyzed by cytofluorometry for CD4 and CD8 expression.

3 Results

3.1 In vitro differentiation of CDS' thymocytes

Total thymocytes were treated twice with anti-CD4 antibody plus complement to deplete the majority of CDCbearing cells and CD8' cells were isolated from this population by panning on anti-CD8-coated dishes (see Sect. 2.3). CD8' thymocytes isolated in this fashion contain a CD8" population, which appears DN by cytofluorometric analysis with antLCD8 and anti-CD4 reagents, but differs from true DN thymocytes in several phenotypic and functional aspects [9]. For example, they adhere specifically to anti-CD8-coated dishes and are depleted of cells expressing the IL2R at the surface [9].

The majority of CD8' thymocytes prepared as above (- 65%) show a high level of CD4 expression after 16 h of culture in regular medium and appear as DP cells (Fig. 1A and B). By 33 h of culture three distinct cell populations were evident (Fig. 1C): (a) DP cells were the major population, (b) CDShi cells which are likely to represent mature CDShi thymocytes present in the starting population and (c) a subset of CD8" thymocytes present within the total starting CD8' population which cannot convert to high CD4 expression (see also Fig. 5 ) . The survival of the CD8' cells in culture was 95% and 48% of starting numbers at 16 and 33 h, respectively. This finding rules out the differential survival of contaminating cells as an explanation for the observed phenotypic changes.

+-----+I------ 10' 10' 10' 10' 10' 10'

CD4 Fluorescence (GK1.5PE)

Figure 1. In vitro conversion of CD8+ thymocytes. CDX' thymocytes from BALB/c mice were isolated as described in Sect. 2.3 and cultured in FCS-containing medium for zero (A), 16 h (B) or 33 h (C). Cyto- fluorometric analysis was carried out to reveal simultaneously CD8 and CD4 expression. At the onset of cultures 42% of CD8' cells were CDS'", and 1.1% of CDX' cells expressed IL 2R (as detected by FITC- PC61 mAb).

3.2 In vitro differentiation of DN thymocytes

DN thymocytes and the HSAg' IL2R' DN subset have been shown to generate DP and CD4' progeny following in vivo intrathymic transfer [4. 6,7]. We therefore wanted to compare the in vitro differentiation potential of DN cells with that of CD8' thymocytes. We prepared DN thymocytes by depleting thymocytes of CD4 bearing cells using antibody plus comple- ment and negatively selecting the surviving population on anti- CD8-coated plates, i . e . the anti-CD8 nonadherent cells were used. This protocol efficiently removes cells expressing even low levels of CD8, allowing us to assess the conversion capac- ity of "true" DN cells.

Fig. 2 shows that, unlike CD8' cells, DN thymocytes are incapable of significant CD8 or CD4 expression changes dur- ing short-term in vitro culture. The recovery of DN cells in

Eur. J. Immunol. 1989.19: 649-653 In vitro differentiation of immature thymocytes 651

0 33

10'

C04 Fluorescence (GKl.5.PE)

Figure 2 . True DN thymocytes do not undergo in vitro differentia- tion. DN thymocytes were prepared from BALB/c thymuses by cytotoxic depletion of CD4' cells followed by depletion of CD8' cells by panning. The resulting population (A) was highly (68.9%) positive for IL2R expression. Cytofluorometric detection of CD8 and CD4 was performed at 0 h (A), 16 h (B) and 33 h (C) of culture.

culture closely paralleled that of CD8+ thymocytes, being 98% and 51% at 16 and 33 h of culture, respectively. Longer cul- ture times, up to 80 h, also failed to reveal any phenotypic changes. It is worth mentioning that within anti-CD4 plus complement-treated thymocytes, all in vitro conversion capac- ity segregated with the CD8+ cells since the results presented in Figs. 1 and 2 come from the same initial population of CD4- depleted cells which was split on the anti-CD8 panning step into CD8+ and CD8- (DN) fractions.

3.3 CD8' thymocytes express low levels of the CD4 molecules

We routinely used two cycles of anti-CD4 antibody plus com- plement treatment to deplete CD4-expressing thymocytes. This treatment consistently spared 2-4% of total thymocytes. Staining the surviving thymocytes with a secondary antibody (FITC-GaRIg) which can reveal cells coated with the rat anti- CD4 antibody indicated that all CD4hi cells had in fact been eliminated. Thus, the highest maximal fluorescence observed in over 20 experiments by this staining protocol was 10.8 arbi- trary units while the mean fluorescence for untreated thymo- cytes stained with anti-CD4 followed by FITC-GaRIg is over 100. However, a discrete shift to higher mean fluorescence values of the entire population was observed after anti-CD4 plus complement treatment (e.g. control 4.8, plus FITC- GaRIg 7.3), leading to the conclusion that some of the anti- CD4 plus complement-treated cells may be coated with low levels of the anti-CD4 antibody. When the anti-CD4 plus com- plement-treated cells were split into anti-CD8 adherent and antLCD8 nonadherent fractions by panning, the adherent population had a higher mean fluorescence (7.2) detected by FITC-GaRIg than the nonadherent population (5.2). [It is important to stress that panning itself does not leave any anti- body coating the recovered adherent or nonadherent cells. For instance, untreated thymocytes after adhering to anti-CD8 plates display a mean fluorescence essentially identical to that of the initial population or the nonadherent fraction (i.e. 3.8 untreated, 3.9 adherent and 3.8 nonadherent) when stained with FITC-GaRIg] ([9] and data not shown).

These observations implied the existence of a CD4'" thymo- cyte population which co-purified with CD8' cells during posi- tive selection for CD8 expression by panning. In order to address this question directly, we performed parallel panning of anti-CD4 plus complement-depleted thymocytes on anti- CD8-coated plates or on GaRIg-coated plates. Cells adherent

aCO4 + C' aCO4 + C' GarigAdherent oCD&Adherent

103

m (P

al u c al a C 0 4 + C' Cells From

oCO&Adherent C. GarigAdherent

1 0 3 1 ''' u M

=! U - co 0 u

10' 10' 10' 101 102 1 0 3

CD4 Fluorescence (GK1.5.PE)

Figure 3. CD8 "single-positive'' thymocytes express a low density of CD4. In Exp. 1 (top panels), anti-CD4 plus complement-depleted C57BL16 thymocytes were positively selected for adherence to GaRIg- coated plates (A) or anti-CD8-coated plates (B). In Exp. 2 (bottom panels) anti-CD4 plus complement-treated thymocytes were selected for adherence to anti-CD8-coated plates (C), and the anti-CD8 ad- herent cells were further selected for adherence to GaRIg-coated plates (D). Percent of cells which were CD8'" (i.e. scoring as DN) was 12.7%. 20.1%, 33.6% and 29.4% for panels A, B, C and D, respec- tively. In no case did the percent of IL2R+ cells exceed 1.5%. FACS analysis was performed to reveal CD8 and CD4 expression.

to GaRIg-coated plates (CD4") display a strikingly similar CD8 cytofluorometric profile to that of anti-CD8 adherent cells (Figs. 3A and B). A considerable fraction of both posi- tively selected populations (up to 20%) appeared DN, yet <2% of the cells expressed the IL2R (in accordance with previous data for CD8+ cells, [9]). The total number of anti- CDCkilled thymocytes which adhered to the plates was 68% (GaRIg) and 79% (anti-CD8) of the input numbers, for panels (A) and (B), respectively.

V Ii 1 0 1 102 1 0 3 I 1 0 1 1 0 2 103

C04 Fluorescence (GK 1 .5 PE)

Figure 4. CD4'" thymocytes convert to DP during in vifro culture. The anti-CD4 plus complement-depleted thymocytes were selected for adherence to GaRIg-coated plates (CD4'", A) or to anti-CD8-coated plates (CD8+, B). Their CD4/CD8 profile before culture is shown in Figs. 3A and B and their profile after 15 h of culture in regular medium is shown here, in (A and B), respectively.

652 J. NikoliC-ZugiC, M. W. Moore and M. J. Bevan

These results indicated that CD8' cells might actually express low levels of CD4. To directly assess the relationship of CD8' and CD4" cells sequential panning steps were performed. Anti-CD4 plus complement-killed thymocytes which had been selected for adherence to anti-CD8-coated plates were then panned on GaRIg-coted plates. Over 70% of CD8' cells adhered to the GaRIg-coated plates, and their cytofluoromet- ric profile was virtually unchanged after the second panning step (Figs. 3C and D).

Eur. J. Immunol. 1989.19: 649-653

Together these results argue strongly that most, and perhaps all, CD8' cells are also CD4'". Further support for this comes from in vitro behavior of CD4" and CD8+ cells. The GaRIg- adherent and anti-CD8-adherent subsets of anti-CD4 killed thymocytes shown in Figs. 3A and B , were cultured overnight and analyzed for CD4 and CD8 expression. Both populations converted to DP with identical kinetics, giving 58% and 56% DP after overnight culture, respectively (Fig. 4).

3.4 In vitro differentiation of CD8' subsets

We have previously shown that CD8' thymocytes can be sepa- rated into CD8hi and CD8l" subsets, both of which possess precursor activity in vivo following intrathymic injection [9]. CD8+CD4'" cells, isolated by anti-CD4 lysis and anti-CD8 panning, were stained with FITC-anti-CD8 and sorted to obtain CD8hi and CD8" subsets. Fig. 5 presents the results of the phenotypic analysis of the sorted subsets before and after overnight culture. Twenty-four percent of CD8hiCD4'0 cells elevated their CD4 expression to high levels (Figs. 5A and B). This is consistent with earlier reports on the percentage of immature CDghi thymocytes among total CD8' thymocytes

10' 102 103 101 10' 103

CD4 Fluorescence (GK1.5-PE)

Figure 5. In vitro differentiation of CDgh' and CD8'" thyrnocytes. Thymocytes from BALB.B mice were treated with anti-CD4 plus complement and the survivors selected for adherence to anti-CD8 coated plates. The plate-adherent CD8+ cells were stained with FITC- anti-CDS and sorted to obtain the brightest [CDSh', (A)] and dullest [CD8'", (C)] 30% of the cells. The resulting populations showed less than 2% overlap. Cells were cultured in regular medium for 14 h and reanalyzed for CD8 and CD4 expression (B and D). Before culture less than 0.2% of CDgh' and 3.1% of CD8'" cells expressed the IL2R.

based on the expression of HSAg [3, 161 or CD3 [17]. CD8'"CD4'" cells are 100% HSAg' (data not shown) and appeared DN at the onset of cultures (Fig. 5C) but, unlike true DN, lacked IL2R-bearing cells (compare legends to Figs. 2 and 5). The vast majority of CD8"CD4'" cells became DP after 14 h of in vitro culture (83% in Fig. 5D), a feature which further distinguishes these cells from true DN cells. Cell viabilities in this experiment were > 85% after 16 h in culture for both CDghi and CD8'" populations.

4 Discussion

Thymocytes which lack expression of the CD4 and CD8 acces- sory molecules i.e. DN thymocytes, and, in particular the sub- set of DN which expresses HSAg and IL2R on the surface, are able to proliferate and give rise to differentiated progeny when they are introduced into an irradiated, adoptive thymus [4, 71. We find that preparations of DN containing up to 70% IL2R' cells are not able to express the CD4 and CD8 accessory molecules during short-term tissue culture (Fig. 2). This find- ing may appear to be at odds with previous work showing that adult DN prepared by anti-CD4, anti-CD5 and anti-CD8 lysis [18] or fetal and neonatal thymocytes treated with antLCD8 plus complement [19] differentiated in vitro. We suspect that any conversion to brightly staining DP observed in vitro can be ascribed to cells in the DN preparation which already express a low surface density of the accessory molecules. Thus, we showed that cells which appear DN by FACS analysis yet which adhere to anti-CD8-coated plates to convert into DP in vitro while cells which do not adhere do the plates show no differentiation in vitro. In other experiments we have found that DN thymocytes prepared by complement lysis with anti- CD4 plus anti-CD8 antibodies show some conversion to DP in vitro. However, if the antibody-coated survivors are removed by panning, the nonadherent DN show no conversion. Thus, the stage in thymocyte development at which the DN cell receives the signal(s) to proceed with the differentiation by initiating the expression of CD8 and CD4 is likely to occur after the transient expression of IL2R by DN cells.

The finding that immature (OX44-) CD8' rat thymocytes rapidly express high levels of CD4 to become DP has been reported recently by Patterson and Williams [20]. Similarly, MacDonald et al. reported that CD8'CD3-cells isolated from murine thymus converted rapidly to DP in vitro [17].We con- firmed here that about 30% of CD8hi thymocytes convert to DP during overnight culture (Fig. 5) and this is consistent with the fact that about 30% of total CD8hi thymocytes are imma- ture as determined by HSAg or CD3 expression [3, 16, 171. Our work also demonstrates that a large percentage of thymo- cytes that express a low surface density of CD8 ( i . e . cells which can be panned on anti-CD8 plates, yet do not stain brightly with FITC-anti-CD8) also convert rapidly to DP during short- term in vitro culture (Fip 5 ) . This result further stresses the distinction between CD8" and true DN cells. These cells have also been shown to differ from true DN cells by the virtual lack of IL2R expression, by their inability to home to the thymus following i.v. injection, and their more rapid repopulation kinetics following intrathymic adoptive transfer [9].

We believe that all of the CD8' thymocytes in anti-CD4 plus complement-treated populations which adhere to anti-CD8- coated plates also express a low surface density of the CD4 molecule. They survive two treatments with anti-CD4 plus

Eur. J. Immunol. 1989.19: 649-653 I n vitro differentiation of immature thymocytes 653

complement yet are coated with low levels of the rat IgM antibody and can adhere to GaRIg-coated plates. The thymo- cyte populations which convert to brightly staining DP during short-term culture therefore appear to be either CD8hi,CD4'o or CD8'o,CD4'0. This is in contrast to the conclusions from previous work on the in vitro differentiation of immature CD8' thymocytes [17, 201 that CD4 expression was switched on de novo during the period of culture.

Received November 4, 1988; in revised form December 27, 1988.

5 References

Fowlkes, B. J., Edison, L., Mathieson, B. J. and Chused, T. M., J. Exp. Med. 1985. 162: 802. Husmann, L. A., Shimonkevitz, R. P., Crispe, I. N. and Bevan, M. J., J. Irnrnunol. 1988. 141: 736. Crispe, I. N. and Bevan, M. J., J. Immunol. 1987. 138: 2013. Scollay, R., Wilson, A., D'Amico, A., Kelly, K., Egerton, M., Pearse, M., Wu, L. and Shortman, K., Irnrnunol. Rev. 1988.104: 81. MacDonald, H. R., Howe, R. C., Pedrazzini, T., Lees, R. K., Budd, R. C., Schneider, R., Liao, N. S., Zinkernagel, R. M., Louis, J. A., Raulet, D. H., Hengartner, H. and Miescher, G., Irnrnunol. Rev. 1988. 104: 157. Crispe, I. N., Moore, M. W., Husmann, L. A., Smith, L., Bevan, M. J. and Shimonkevitz, R. P., Nature 1987. 329: 336.

7 Shimonkevitz, R. P., Husmann, L. A., Bevan, M. J. andcrispe, I. N., Nature 1987. 329: 157.

8 Kisielow, P., Leisserson, W. and Von Boehmer, H., J. Immunol. 1984. 13:: 1117.

9 NikoliC-ZugiC, J. and Bevan, M. J., Proc. Natl. Acad. Sci. USA 1988. 85: 8633.

10 Ceredig, R., Dialynas, D. P., Fitch, F. W. and MacDonald, H. R., J. Exp. Med. 1983. 158: 1654.

11 Sarmiento, M., Glasebrook, A. L. and Fitch, F. W., J. Imrnunol. 1980. 125: 2665.

12 Ceredig, R., Lowenthal, J. W., Nabholz, M. and MacDonald, H. R., Nature 1985. 314: 98.

13 Dialynas, D. P., Quan, Z. S., Wall, K. A., Pierres, A., Quintans, J., Loken, M., Pierres, M. and Fitch, F. W., J. Immunol. 1983. 131: 2445.

14 Ledbetter, J. A. and Herzenberg, L. A., Immunol. Rev. 1979. 75: 2844.

15 Wysocki, L. J. and Sato, V. L., Proc. Natl. Acad. Sci. USA 1978. 75: 2844.

16 Shortman, K.,Wilson, A., Egerton, M., Pearse, M. and Scollay, R., Cell. lmrnunol. 1988.113: 462.

17 MacDonald, H. R., Budd, R. C. and Howe, R. C., Eur. J. lmmu- nol. 1988. 18: 519.

18 Fowlkes, B. J . , Edison, L., Mathieson, B. J. and Chused, T. M., in Cantor, H., Chess, L. and Sercarz, E. (Eds.), The regulation of immune system, UCLA Symp. MoL Cell. Biol. New Ser. 1984. 18: 275.

19 Ceredig, R., Sekaly, R. P. and MacDonald, H. R., Nature 1983. 303: 248.

20 Paterson, D. J . and Williams, A. F., J. Exp. Med. 1987. 166: 1603.