Eur. J. Immunol. 1990.20: 1061-1066 Mikael Andersson and Rikard Holmdahl Department of Medical and Physiological Chemistry, Uppsala University, Uppsala
Type I1 collagen-reactive Tcells in the mouse
1061
Analysis of type I1 collagen-reactive T cells in the mouse I. Different regulation of autoreactive vs. non-autoreactive anti-type I1 collagen T cells in the DBA/l mouse* -
The T cell reactivity against type II collagen (CII) was analyzed in the collagen-induced arthritis-susceptible mouse strain DBA/l. It was shown that the proliferative response in lymph node cells from rat CII-immunized mice was mainly directed against a foreign determinant present on all heterologous CII tested but not on autologous CII. A Tcell line with this reactivity reacted with high sensitivity with CII and the determinant was mapped to the C B l l fragment of CII. A weak autoreactive response could be detected in the primary cultures using high concentrations of mouse CII and this reactivity remained after several stimulations with high concentrations of rat CII but not with low concentrations of rat CII. A similar response against mouse CII but with only limited cross-reactivity to rat CII was seen when culturing the cells with mouse CII as antigen. The optimal concentration for the autoreactive response was always more than 100-fold higher than for the response of the T cells specific for heterologous CII. An anti-CII Tcell response could also be detected in spleen cells from unimmunized mice and the strongest response was obtained using autologous CII.These results suggest that Tcells recognizing self CII are normally activated in the DBA/1 mouse and possibly as a consequence exhibit a clonal anergy pattern with a weak proliferative response only at high concentrations of
cn .
1 Introduction Immunization with type I1 collagen (CII) leads to development of arthritis (collagen-induced arthritis, CIA), in mice of certain MHC haplotypes and is associated with an immune response against CII [ l , 21. This immune response is Tcell dependent [3] thus Tcells play a crucial role in the induction phase of the disease. The role of Tcells in the effector phase is unclear; however, transfer of the disease with T cells [4] and immunohistochemical detection of activated T cells in early arthritis [5] suggest an important role. Antibodies have also been shown to be involved as indicated by the fact that anti-CII serum can transfer disease [6]. In the arthritis-susceptible mouse strain DBA/1 (H-2‘9 immunization with heterologous CII (rat, bovine or chick CII) leads to a strong anti-CII antibody response with a high degree of cross-reactivity against mouse CII and development of an acute and severe arthritis [7-91. After immunization with autologous mouse CII on the other hand, the
[I 81501
*
This work was supported by funds from the Swedish Medical Research Council, Craaford Foundation, King Gustav V’s 80-
years Foundation, Swedish Association Against Rheumatism and Nanna Swartz Foundation. Correspondence: Mikael Andersson, Department of Medical and Physiological Chemistry, Uppsala University, Box 575, S-75123 Uppsala, Sweden
Abbreviations: CII: Type I1 collagen SC: Spleen cell 0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1990
antibody response against CII is lower with a large individual variation and the disease is more chronic with a later onset and lower incidence [7,10,11].These findings suggest that immunogenicity differs between heterologous and autologous CII but also that Tcells capable of recognizing autologous CII must exist. When analyzing different mouse strains for susceptibility to CIA and immune response against CII we could show that only strains developing antibodies after immunization with autologous CII were capable of developing arthritis after immunization with autologous or heterologous CII [ ll] . Some strains developing autoantibodies only after immunization with heterologous CII did not develop arthritis.These results suggest that even in the presence of an anti-CII autoantibody response, autoreactive T cells are essential for the development of disease. Studies on T cells in experimental autoimmune models have mostly involved establishment of autoreactive T cell lines which have been used to transfer the disease, such as in experimental autoimmune encephalomyelitis [ 121.We have used a similar approach for CIA and have shown earlier that CII-reactive T cells can transfer disease t o naive DBAA mice [4]. The transfer was, however, only effective when using a T cell h e , and a specific clone responsible for disease induction could not be isolated [13]. Another line of research on autoimmunity has involved the detection of autoreactive T cells in naive mice. Coutinho et al. have suggested that autoreactivity is a common property of the “normal1y”activated T cells [14]. Hooper et al. showed that mice harbor T cells reactive with mouse erythrocytes [15]. The fact that these mice remain healthy shows that tolerance can be maintained in the presence of autoreactive Tcells and of an autoantigen that is exposed to the immune system. OO14-2980/~0/05051061$02.50/0
1062
Eur. J. Immunol. 1990. 20: 1061-1066
M. Anderson and R. Holmdahl
This report presents data on the T cell reactivity against CII in immunized and unprimed DBA/1 mice. We show that T cells specific for one or more heterologous determinant(s) on CII as well asTcells cross-reacting with autologous CII are present in the DBA/l mouse and that theseT cells show a different pattern of behavior. We also show that pepsin, which may be present as a contaminant in collagen preparations, is a strongT cell antigen and that this must be taken into consideration when analyzing theT cell response against CII.
the LN cells cultured in triplicates in flat-bottom 96-well plates (Nunc, Roskilde, Denmark) in AS medium at a concentration of 5 x lo6cells/ml for 96 h followed by 16 h in the presence of 1 pCi (= 37 kBq) of [3H]dThd (Amersham Int., Amersham, Bucks., GB).T cell lines were cultured at a concentration of 2 x 105-3 x 105 cells/ml together with 2.5 x lo6 irradiated (2000 rad) syngeneic SC/ml for 48 h followed by a 16-h pulse with [3H]dThd. 2.4 Establishment of T cell lines
2 Materials and methods 2.1 Animals and collagens DBA/1 mice, originally obtained from Jackson Laboratories (Bar Harbor, ME), were kept and bred at the Biomedical Center, Uppsala, Sweden. CII was prepared from a rat chondrosarcoma [161, mouse xiphoideus, bovine nasal cartilage and human fetal sternum. Type I collagen was prepared from rat skin by pepsin digestion as described [171.The cartilage was treated in 4 M guanidinium chloride to remove proteoglycans. CII was solubilized with pepsin (Worthington, Freehold, NJ) in 0.5 M acetic acid. Collagen was precipitated from the solubilized cartilage by addition of NaCl to a final concentration of 2 M. After dialysis against neutral buffer the collagen was passed through a column of DEAE-Sepharose 6B (Pharmacia Biotechnology, Uppsala, Sweden). This step was shown to be very important since most of the pepsin bound to the column. The minor cartilage collagens were removed by precipitation at a concentration of 0.7 M NaCl in 0.5 M acetic acid. Type I collagen was removed by precipitation at 2.5 M NaCl in neutral buffer. Lathyritic CII was prepared from chondrosarcoma grown in rats made lathyritic by treatment with proprionitrile fumarate. In this case the collagen was solubilized by neutral salt extraction followed by salt precipitation and thereafter the purification was continued as described above. Chick CII was a generous gift from Dr. Klaus von der Mark, Erlangen, FRG. CNBr peptides derived from chick CII [18] were a kind gift from Dr. Michael Cremer (Memphis,TN) .The collagens were stored freeze-dried and dissolved in 0.1 M acetic acid before use. Denaturation was performed by treating the collagen at 50°C for 30 min. 2.2 Culture media DMEM supplemented with Hepes, L-glutamine, penicillin and streptomycin was used throughout. Medium used for antigen stimulation (AS medium) was supplemented with 1% fresh normal mouse serum. Medium used for propagation of Tcell lines (Tmedium) was supplemented with 8% FCS (KC Biological, Lenexa, KS) and an optimal amount of SN from Con A-stimulated rat spleen cells (SC). 2.3 Proliferation assays The proliferative response of LN cells after immunization with CII was measured as described [4] with minor modifications. Briefly, 7-12 days after immunization with CII in the hind footpads the draining LN were removed and
T cell lines were established from primary cultures of LN cells from CII-immunized mice as described earlier [4]. The LN cells were cultured with denatured rat CII at a concentration of 25-100 pg/ml. After 96 h the cells were resuspended in T medium which was changed every 2nd or 3rd day. TheT cells were restimulated every 10th to 20th day with 1-50 yglml denatured rat CII in the presence of 2.5 x lo6 irradiated SC/ml. 2.5 T cell cultures from spleens of naive mice Tcell reactivity against CII was analyzed in SC from naive mice as adopted from Hooper et al. [15]. SC were treated with NH&I to remove erythrocytes and cultured in AS medium at a concentration of 5 X 106/ml in flat-bottom 96-well plates (Nunc). After 6 days 1 pCi of C3H]dThd was added and 16 h later the cells were harvested onto glass fiber filters. The anti-CD4 mAb GK1.5 [19] was used for inhibition studies. Bulk cultures of SC were set up in 24-well plates (Nunc) with 50 pg/ml mouse CII. The cells were resuspended in T medium after 6days, expanded for 1 week and then analyzed in a proliferation assay as described above.
3 Results 3.1 Proliferative response of LN cells from CII-immunized mice DBA/1 mice were immunized with 100 pg rat CII emulsified in CFA and 7 days later the draining LN cells were cultured with rat and mouse CII and CNBr fragments of CII from chick and rat. As shown in Fig. 1 we could detect a strong response against rat CII but no response against mouse CII. The response was directed against the C B l l fragment of CII. The absence of autoreactivity is in contrast to earlier findings [4]. We noticed that using different preparations of CII for immunization or in the cultures led to different results. By further analysis we found that pepsin which is used for preparation of CII, acts as a strongT cell antigen, and we could show that some of the earlier presented Tcell lines and clones contained pepsin-specificT cells. This is exemplified by the response of the Tcell clone DCR 34 which was isolated from CII-immunized mice [4]. From the results shown in Fig. 2 it could be concluded that most of the pepsin could be eliminated by passage of the collagen on a DEAE-Sepharose column. However, small amounts of pepsin present in the preparations could give rise to a T cell response as shown in Fig. 3. Immunization with lathyritic rat CII gave no anti-pepsin response (Fig. 3A) while immuniziation with pepsin-digested mouse
Eur. J. Immunol. 1990.20: 1061-1066 30
v p e I1 collagen-reactiveTcells in the mouse
1063
1
r
0 0
10
1
100
(B)
0.5
5
10
1
0 0
100
wlml
50
wml
Figure 1. Proliferative response of LN cells 7 days after immunization with 100 pg lathyritic rat CII in the hind footpads. Antigens were in (A) rat CII ( 0 )and mouse CII (m) and in (B) CNBr fragments CB8 (0),CBlO (0) and CBll (A) derived from chick CII.
Figure 3. Proliferative response of LN cells 7 days after immunization with 100 pg lathyritic rat CII (A) and 100 pg pepsin-digested mouse CII (B). Antigens were lathyritic rat CII (0),pepsindigested mouse CII).( and pepsin (0).
1201
0
2
10
50
250
Pglml 801
Figure 4. Proliferative response of LN cells 12 days after immunization with 200 pg lathyritic rat CII. Antigens were rat CII (0)and mouse CII (m).
60-
7
s x
40-
E
8
20
o
104 10-3 10-2 1 0 ’ ~ 1
10
100
Pglml
Figure2. Proliferative response of the T cell clone DCR34. Antigens were in (A) lathyritic rat CII (U), pepsin-digestedrat CII not passed on DEAE-Sepharose (0),pepsin-digested rat CII passed on DEAE-Sepharose (m) and in (B) pepsin (A).
whether this response was directed against a mouse CIIspecific epitope or against pepsin. However, using a very high antigen concentration a response against mouse CII could be detected also after immunization with pepsin-free rat CII. In Fig. 4 the response 12 days after immunization with 200 pg lathyritic rat CII is shown. At the highest concentration of CII (250 pg/ml) a proliferative response against mouse CII appeared.
3.2 Analysis of T cell lines CII gave a significant anti-pepsin response (Fig. 3B). To ascertain that pepsin-reactive T cells were not present among the proliferating cells it was therefore necessary to immunize mice with lathyritic rat CII prepared without pepsin digestion. Since we could not prepare lathyritic mouse CII the response after immunization with mouse CII could not be clearly analyzed. As shown in Fig. 3B there was no response against lathyritic rat CII after immunization with mouse CII while there was a significant response against mouse CII. It was not possible to determine
Since the autoreactivity in the primary cultures was only detected with high concentrations of CII, LN cells from mice immunized with lathyritic rat CII were repeatedly stimulated with high concentrations (50-100 pg/ml) or low concentrations (1-5 pg/ml) of rat CII. The response after four such passages is shown in Fig. 5. Both these lines reacted primarily with rat CII with high sensitivity and a high stimulation index (SI, cpm with antigen/cpm of medium control), SI = 130 at 0.4 p g / d for the line cultured with high concentrations of rat CII. The line cultured with
Eur. J. Immunol. 1990. 20: 1061-1066
M. Andersson and R. Holmdahl
1064
2oo
? ? E
1
100
4 0
0
0.01
0.1
1
10
100
Clslml
(B)
301
' 0
0.0180.08 O 0.4
0
2 O 10
50
250 L
ouc% 0
0.4
2
10
50
250
w m l
Calm1
Figure5. Proliferative response of T cells derived from mice immunized with rat CII. After a primary culture of the LN cells with rat CII at 100 pg/ml (A, B) or 25 pglml (C, D) the cells were stimulated on four consecutive occasions with 50 pg/ml of rat CII (A, B) or 1-5 pglml of rat CII (C, D) and then assayed. Antigens were rat CII ( 0 )and mouse CII (W). In (B) and (D) is shown the response against mouse CII at a higher concentration range only.
6ol
4g20j01 & f
? ?
20-
X
;
10-
0
0
0.025
0.25
2.5
25
nglrnl
Figure 7. Proliferative response of the T cell line M1. Antigens tested were: in (A) denatured rat CII (0),denatured bovine CII (W), native rat CII (0),native bovine CII ( O ) , native chick CII (A), native human CII (A), native mouse CII (0)and rat collagen I (V) and in (B) CNBr fragments CB8 (0),CBlO (W), CBll (0)and CB12 ( 0 )derived from chick CII.
50
4-
?i
(A)
30
10 "
.
.
.
.
0 0.0180.08 0.4
.
.
2
10
I
50
I
250
3-
? Y
2-
ti
Wml
Figure6. Proliferative response of T cells derived from mice immunized with rat CII. After a primary culture of the L N cells with mouse CII at 100 pglml the cells were repeatedly stimulated with 50 pglml of mouse CII. The assay was performed after six cycles of antigen stimulations. Antigens were rat CII (0) and mouse CII (W).
5-
(8)
4-
?
z
3-
Y
E, 2 0
high concentrations of CII still showed autoreactivity (SI = 6 at 50 pg/ml) while autoreactivity was not visible in the line cultured with low concentrations of CII. On the other hand, the same LN cells from mice immunized with lathyritic rat CII when repeatedly stimulated with 50-100 pg/ml mouse CII yielded a specific response primarily against mouse CII as shown in Fig. 6. The concentration needed and the SI were similar to those found for the autoreactivity when culturing the cells with high concentrations of rat CII (SI = 5 at 50 pg/ml). In this case the reactivity against rat CII was weaker than the autoreactive response. Several T cell lines were established from mice immunized with rat CII. None of them showed any autoreactivity. These lines were cultured with rat CII concentrations giving optimal proliferation (1-5 pg/ml).TheT cell line M1 which
1-
0-
I
was established in this manner was used for more thorough specificity analysis. After more than five cycles of expansion of T cells inT medium followed by antigen stimulation with rat CII and APC the proliferative response was analyzed. The results are shown in Fig. 7. The Tcell line
Eur. J. Immunol. 1990.20: 1061-1066
3001
/
q p e I1 collagen-reactive T cells in the mouse
1065
globulin gives a much stronger T cell response than autologous thyroglobulin [26]. It was also shown that a T cell line which did not proliferate in response to autologous thyroglobulin could induce thyroiditis upon transfer [27].
We have been working on the CIA model using the same approach. Earlier we presented data on CII-reactive Tcell lines capable of transferring disease to naive recipients [4]. An arthritogenic clone could not, however, be isolated. 0’ 0 1 10 100 After further characterizing the T cell response against CII figlml in DBA/l mice we have now shown that there is a Figure 9. Proliferative response in a secondary culture of SC from pronounced difference in the response against heterolonaive DBA/1 mice. The SC were cultured with mouse CII at a gous vs. autologous CII. Upon immunization with heteroconcentration of 50 pg/ml and then expanded in T medium. The logous CII, we could isolate T cells which proliferated in blasts were then assayed in a proliferation assay with rat CII (0) response to all heterologous CII tested at concentrations and mouse CII (m). < 0.1 pg/ml but not to mouse CII. The presence of autoreactive T cells could be demonstrated but these cells showed only a weak proliferation and only at high concenresponded strongly to rat, bovine, human and chick CII but trations of CII. We were able to show that the earlier Tcell not at all to mouse CII (Fig. 7A). Denatured CII gave a lines and clones described as autoreactive contained Tcells somewhat better response than native CII. No response was reacting with pepsin and we believe that this reactivity was seen with rat type I collagen. The proliferative response responsible for most of the reactivity against mouse CII. against different CNBr peptides derived from chick CII is However, even in the absence of proliferation against shown in Fig. 7B. The response was directed against the autologous CII autoreactiveTcells may be present in the T CBll peptide. cell lines as bystander cells only capable of proliferating if other cells producing the relevant growth factors are present. Such a possibility would be in accordance with the 3.3 T cell response of SC from naive mice in vitro model of Tcell tolerance described by Jenkins et al., where the tolerant T cell could be activated by its antigen As described by others [15] autoreactive T cells can be and express IL 2R but was incapable of producing IL 2 and detected in spleens of naive mice.We therefore analyzed SC therefore did not proliferate [28]. This might also account from naive DBA/1 mice for CII reactivity. The SC were for (a) the transfer of thyroiditis with aTcell line incapable cultured for 6 days, pulsed with [3H]dThd and harvested. of proliferating in response to autologous thyroglobulin The results of such an assay are shown in Fig. 8A. A [27] and (b) the fact that autologous MBP could trigger L N significant response against mouse CII and rat CII but not cells from MPB-primed rats to transfer encephalomyelitis rat type1 collagen was found. The response could be in the absence of a proliferative response [29]. Another inhibited by the addition of 10 pg/ml of the anti-CD4 mAb feature of the Jenkins model for clonal anergy which fits GK1.5 (Fig. 8B). To further verify the Tcell nature of the with our data is that the unresponsiveness to antigen can be response, SC cultured with 50 pg/ml of mouse CII were at least partly overcome with a high amount of antigen expanded in Tmedium and then assayed. As seen in Fig. 9 ~301. these cells showed an increased response against mouse CII and rat CII. The presence of two types of T cells may explain the difference in immunogenicity between heterologous and autologous CII. Immunization with heterologous CII gives rise to aTcell response primarily directed against a foreign 4 Discussion determinant on the CII molecule.TheseTcells interact with Self-nonself discrimination in the immune system is still a cross-reactive B cells [7] inducing a strong autoantibody matter of debate (reviewed in [20]). Both negative selection response. We were able t o show that this reactivity was of T cells based on self recognition in the thymus [21-231 as directed primarily against the C B l l fragment. This fragwell as peripheral tolerance in the absence of deletion [24] ment has earlier been shown to be the most immunogenic have been demonstrated. It would be reasonable to assume part of the CII molecule [31]. Immunization with autolothat these two mechanisms complement each other with gous CII on the other hand may trigger the autoreactive clonal deletion in the thymus acting on T cells recognizing anti-CII Tcell but since these cells show a weaker response ubiquitous antigens and peripheral tolerance acting on T the antibody response becomes much weaker [8,9]. cells recognizing more unevenly distributed antigens. If the T cells recognizing autologous CII in the DBA/l Autoreactive T cells have been shown t o be involved in mouse actually exhibit the type of clonal anergy discussed several experimental autoimmune disease models. The earlier this means that they have already interacted with most studied model in this respect is the experimental their antigen in the peripheral immune system of the naive autoimmune encephalomyelitis model. Autoreactive T cell mouse. This in turn suggests that it should actually be clones are readily isolated from rats and mice immunized possible t o detect them without previous immunization. In with myelin basic protein (MBP) and these T cells will a study of Hooper et al. [15] it was shown that Tcells from induce disaese [12] upon transfer to naive recipients. In the spleens of unprimed mice would proliferate in response experimental autoimmune thyroiditis similar data have to autologous erythrocytes. Using a similar protocol we been produced [25] but in this model heterologous thyro- were able to show proliferation against autologous CII in 1
1066
M. Andersson and R. Holmdahl
SC from DBA/1 mice. The response could be inhibited by the anti-CD4 antibody GK1.5. Although significant, the response against rat CII was not as strong as that against autologous CII. The proliferating cells were expanded in ILZcontaining medium and in a secondary culture an enrichment of autoreactiveTcells could be demonstrated. From the fact that in the unimmunized mouse the response was stronger against autologous CII than against heterologous CII we conclude that the autoreactive cells must have been activated in vivo. Based on the data presented here we would like to propose the followinghypothesis: autoreactive anti-CII T cells exist in the DBA/1 mouse which are not deleted but are present in the periphery. These T cells are subject to induction of peripheral tolerance and cannot be readily induced to proliferate. The fact that they can be detected as proliferating cells in cultures from spleens of unprimed mice indicate that activation is a step in the tolerance induction. However, the tolerance is not absolute and the Tcells can upon immunization be induced to perform certain effector functions, such as B cell help and DTH reactions, and thereby participate in the induction of arthritis. Immunization with heterologous CII will also stimulate T cells which are not autoreactive and therefore not regulated. Activation of these T cells might further enhance the activation of the autoreactive T cells. Further analysis of these two types of Tcells will be the subject of our future work. Received November 30, 1989; in revised form January 15, 1990.
5 References 1 Courtenay, J. S., Dallman, M. J., Dayan, A. D., Martin, A. and Mosedale, B., Nature 1980. 283: 666. 2 Wooley, F! H., Luthra, H. S., Stuart, J. M. and David, C. S., J. Exp. Med. 1981. 154: 688. 3 Ranges, G. E., Sriram, S. and Cooper, S. M., J. Exp. Med. 1985.162: 1105. 4 Holmdahl, R., Klareskog, L., Rubin, K., Larsson, E. and Wigzell, H., Scand. J. Immunol. 1985. 22: 295. 5 Holmdahl, R., Jonsson, R., Larsson, F! and Klareskog, L., Lab. Invest. 1988. 58: 53. 6 Stuart, J. M. and Dixon, F. J., J. Exp. Med. 1983.158: 378.
Eur. J. Immunol. 1990. 20: 1061-1066
7 Holmdahl, R., Jansson, L., Gullberg, D., Rubin, K., Forsberg, l? 0. and Klareskog, L., Clin. Exp. Immunol. 1985. 62: 639. 8 Holmdahl, R., Klareskog, L., Andersson, M. and Hansen, C, Immunogenetics 1986. 24: 84. 9 Holmdahl, R., Jansson, L., Andersson, M. and Larsson, E., Immunology 1988. 65: 305. 10 Holmdahl, R., Jansson, L., Larsson, E., Rubin, K. and Klareskog, L., Arthritis Rheum. 1986. 29: 106. Carlioz, A., Roudier, R. and 11 Boissier, M.-C., Feng, X.-Z., Fournier, C., Ann. Rheum. Dis. 1987. 46: 691. 12 Ben-Nun, A.,Wekerle, H. and Cohen, I. R., Eur. J. lmmunol. 1981. 11: 195. 13 Holmdahl R., Andersson, M., Enander, I., Goldschmidt,T., Jansson, L., Larsson, F!, Mo, J., Nordling, C. and Klareskog, L., Int. Rev. Immunol. 1988. 4: 49. 14 Coutinho, A., Marquez, C., Araujo, F? M. F., Pereira, I?, Toribio, M. L., Marcos, M. A. R. andMartinez-A., C., Eur. J. Immunol. 1987. 17: 821. 15 Hooper, D. G. and Taylor, R. B., Eur. J. Immunol. 1987. 17: 797. 16 Smith, B. D., Martin, G. R., Miller, E. J., Dorfman, A. and Swarm, R., Arch. Biochem. Biophys. 1975.166: 181. 17 Miller, E. J. and Rhodes, R. K., Methoak Enzymol. 1982.82: 33. 18 Terato, K., Cremer, M. A., Hasty, K. A., Kang, A. H., Hasty, D. L. and Townes, A. S., Collagen Rel. Res. 1985. 5: 469. 19 Dialynas, D. F! ,Wilde, D. B., Marrack, F!,Pierres, A., Wall, K. A., Havran,W. L., Otten, G. R., Loken, M. R., Pierres, M., Kappler, J. W. and Fitch, E W., J. lmmunol. 1983. 131: 2445. 20 Schwartz, R. H., Cell 1989. 57: 1073. 21 Kappler, J. W., Roehm, N. and Marrack, F!, Cell 1987. 49: 273. 22 MacDonald, H R., Schneider, R., Lees, R. K., Howe, R. C., Acha-Orbea, H., Festenstein, H., Zinkernagel, R. M. and Hengartner, H., Nature 1988. 332: 40. 23 Kisielow, F!, Bliithmann, H., Staerz, U. D., Steinmetz, M. and Von Boehmer, H., Nature 1988. 333: 742. 24 Burkly, L. C., Lo, D., Kanagawa, O., Brinster, R. L. and Flavell, R. A., Nature 1989.342: 564. 25 Romball, C. G. and Weigle, W. O., J. Immunol. 1987. 138: 1092. 26 Romball, C. G. and Weigle,W. O., Eur. J. Immunol. 1984.14: 887. 27 Maron, R., Zerubavel, R., Friedman, A. and Cohen, I. R., J. Immunol. 1983. 131: 2316. 28 Jenkins, M. K. and Schwartz, R. H., J. Exp. Med. 1987. 165: 302. 29 Hunter, S., Cell. Immunol. 1986. 97: 204. 30 Jenkins, M. K., Ashwell, J. D. and Schwartz, R. H., J. lmmunol. 1988.140: 3324. 31 Terato, K., Hasty, K. A., Cremer, M. A., Stuart, J. M.,Townes, A. S. and Kang, A. H., J. Exp. Med. 1985. 162: 637.
Type I1 collagen-reactive Tcells in the mouse
1061
Analysis of type I1 collagen-reactive T cells in the mouse I. Different regulation of autoreactive vs. non-autoreactive anti-type I1 collagen T cells in the DBA/l mouse* -
The T cell reactivity against type II collagen (CII) was analyzed in the collagen-induced arthritis-susceptible mouse strain DBA/l. It was shown that the proliferative response in lymph node cells from rat CII-immunized mice was mainly directed against a foreign determinant present on all heterologous CII tested but not on autologous CII. A Tcell line with this reactivity reacted with high sensitivity with CII and the determinant was mapped to the C B l l fragment of CII. A weak autoreactive response could be detected in the primary cultures using high concentrations of mouse CII and this reactivity remained after several stimulations with high concentrations of rat CII but not with low concentrations of rat CII. A similar response against mouse CII but with only limited cross-reactivity to rat CII was seen when culturing the cells with mouse CII as antigen. The optimal concentration for the autoreactive response was always more than 100-fold higher than for the response of the T cells specific for heterologous CII. An anti-CII Tcell response could also be detected in spleen cells from unimmunized mice and the strongest response was obtained using autologous CII.These results suggest that Tcells recognizing self CII are normally activated in the DBA/1 mouse and possibly as a consequence exhibit a clonal anergy pattern with a weak proliferative response only at high concentrations of
cn .
1 Introduction Immunization with type I1 collagen (CII) leads to development of arthritis (collagen-induced arthritis, CIA), in mice of certain MHC haplotypes and is associated with an immune response against CII [ l , 21. This immune response is Tcell dependent [3] thus Tcells play a crucial role in the induction phase of the disease. The role of Tcells in the effector phase is unclear; however, transfer of the disease with T cells [4] and immunohistochemical detection of activated T cells in early arthritis [5] suggest an important role. Antibodies have also been shown to be involved as indicated by the fact that anti-CII serum can transfer disease [6]. In the arthritis-susceptible mouse strain DBA/1 (H-2‘9 immunization with heterologous CII (rat, bovine or chick CII) leads to a strong anti-CII antibody response with a high degree of cross-reactivity against mouse CII and development of an acute and severe arthritis [7-91. After immunization with autologous mouse CII on the other hand, the
[I 81501
*
This work was supported by funds from the Swedish Medical Research Council, Craaford Foundation, King Gustav V’s 80-
years Foundation, Swedish Association Against Rheumatism and Nanna Swartz Foundation. Correspondence: Mikael Andersson, Department of Medical and Physiological Chemistry, Uppsala University, Box 575, S-75123 Uppsala, Sweden
Abbreviations: CII: Type I1 collagen SC: Spleen cell 0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1990
antibody response against CII is lower with a large individual variation and the disease is more chronic with a later onset and lower incidence [7,10,11].These findings suggest that immunogenicity differs between heterologous and autologous CII but also that Tcells capable of recognizing autologous CII must exist. When analyzing different mouse strains for susceptibility to CIA and immune response against CII we could show that only strains developing antibodies after immunization with autologous CII were capable of developing arthritis after immunization with autologous or heterologous CII [ ll] . Some strains developing autoantibodies only after immunization with heterologous CII did not develop arthritis.These results suggest that even in the presence of an anti-CII autoantibody response, autoreactive T cells are essential for the development of disease. Studies on T cells in experimental autoimmune models have mostly involved establishment of autoreactive T cell lines which have been used to transfer the disease, such as in experimental autoimmune encephalomyelitis [ 121.We have used a similar approach for CIA and have shown earlier that CII-reactive T cells can transfer disease t o naive DBAA mice [4]. The transfer was, however, only effective when using a T cell h e , and a specific clone responsible for disease induction could not be isolated [13]. Another line of research on autoimmunity has involved the detection of autoreactive T cells in naive mice. Coutinho et al. have suggested that autoreactivity is a common property of the “normal1y”activated T cells [14]. Hooper et al. showed that mice harbor T cells reactive with mouse erythrocytes [15]. The fact that these mice remain healthy shows that tolerance can be maintained in the presence of autoreactive Tcells and of an autoantigen that is exposed to the immune system. OO14-2980/~0/05051061$02.50/0
1062
Eur. J. Immunol. 1990. 20: 1061-1066
M. Anderson and R. Holmdahl
This report presents data on the T cell reactivity against CII in immunized and unprimed DBA/1 mice. We show that T cells specific for one or more heterologous determinant(s) on CII as well asTcells cross-reacting with autologous CII are present in the DBA/l mouse and that theseT cells show a different pattern of behavior. We also show that pepsin, which may be present as a contaminant in collagen preparations, is a strongT cell antigen and that this must be taken into consideration when analyzing theT cell response against CII.
the LN cells cultured in triplicates in flat-bottom 96-well plates (Nunc, Roskilde, Denmark) in AS medium at a concentration of 5 x lo6cells/ml for 96 h followed by 16 h in the presence of 1 pCi (= 37 kBq) of [3H]dThd (Amersham Int., Amersham, Bucks., GB).T cell lines were cultured at a concentration of 2 x 105-3 x 105 cells/ml together with 2.5 x lo6 irradiated (2000 rad) syngeneic SC/ml for 48 h followed by a 16-h pulse with [3H]dThd. 2.4 Establishment of T cell lines
2 Materials and methods 2.1 Animals and collagens DBA/1 mice, originally obtained from Jackson Laboratories (Bar Harbor, ME), were kept and bred at the Biomedical Center, Uppsala, Sweden. CII was prepared from a rat chondrosarcoma [161, mouse xiphoideus, bovine nasal cartilage and human fetal sternum. Type I collagen was prepared from rat skin by pepsin digestion as described [171.The cartilage was treated in 4 M guanidinium chloride to remove proteoglycans. CII was solubilized with pepsin (Worthington, Freehold, NJ) in 0.5 M acetic acid. Collagen was precipitated from the solubilized cartilage by addition of NaCl to a final concentration of 2 M. After dialysis against neutral buffer the collagen was passed through a column of DEAE-Sepharose 6B (Pharmacia Biotechnology, Uppsala, Sweden). This step was shown to be very important since most of the pepsin bound to the column. The minor cartilage collagens were removed by precipitation at a concentration of 0.7 M NaCl in 0.5 M acetic acid. Type I collagen was removed by precipitation at 2.5 M NaCl in neutral buffer. Lathyritic CII was prepared from chondrosarcoma grown in rats made lathyritic by treatment with proprionitrile fumarate. In this case the collagen was solubilized by neutral salt extraction followed by salt precipitation and thereafter the purification was continued as described above. Chick CII was a generous gift from Dr. Klaus von der Mark, Erlangen, FRG. CNBr peptides derived from chick CII [18] were a kind gift from Dr. Michael Cremer (Memphis,TN) .The collagens were stored freeze-dried and dissolved in 0.1 M acetic acid before use. Denaturation was performed by treating the collagen at 50°C for 30 min. 2.2 Culture media DMEM supplemented with Hepes, L-glutamine, penicillin and streptomycin was used throughout. Medium used for antigen stimulation (AS medium) was supplemented with 1% fresh normal mouse serum. Medium used for propagation of Tcell lines (Tmedium) was supplemented with 8% FCS (KC Biological, Lenexa, KS) and an optimal amount of SN from Con A-stimulated rat spleen cells (SC). 2.3 Proliferation assays The proliferative response of LN cells after immunization with CII was measured as described [4] with minor modifications. Briefly, 7-12 days after immunization with CII in the hind footpads the draining LN were removed and
T cell lines were established from primary cultures of LN cells from CII-immunized mice as described earlier [4]. The LN cells were cultured with denatured rat CII at a concentration of 25-100 pg/ml. After 96 h the cells were resuspended in T medium which was changed every 2nd or 3rd day. TheT cells were restimulated every 10th to 20th day with 1-50 yglml denatured rat CII in the presence of 2.5 x lo6 irradiated SC/ml. 2.5 T cell cultures from spleens of naive mice Tcell reactivity against CII was analyzed in SC from naive mice as adopted from Hooper et al. [15]. SC were treated with NH&I to remove erythrocytes and cultured in AS medium at a concentration of 5 X 106/ml in flat-bottom 96-well plates (Nunc). After 6 days 1 pCi of C3H]dThd was added and 16 h later the cells were harvested onto glass fiber filters. The anti-CD4 mAb GK1.5 [19] was used for inhibition studies. Bulk cultures of SC were set up in 24-well plates (Nunc) with 50 pg/ml mouse CII. The cells were resuspended in T medium after 6days, expanded for 1 week and then analyzed in a proliferation assay as described above.
3 Results 3.1 Proliferative response of LN cells from CII-immunized mice DBA/1 mice were immunized with 100 pg rat CII emulsified in CFA and 7 days later the draining LN cells were cultured with rat and mouse CII and CNBr fragments of CII from chick and rat. As shown in Fig. 1 we could detect a strong response against rat CII but no response against mouse CII. The response was directed against the C B l l fragment of CII. The absence of autoreactivity is in contrast to earlier findings [4]. We noticed that using different preparations of CII for immunization or in the cultures led to different results. By further analysis we found that pepsin which is used for preparation of CII, acts as a strongT cell antigen, and we could show that some of the earlier presented Tcell lines and clones contained pepsin-specificT cells. This is exemplified by the response of the Tcell clone DCR 34 which was isolated from CII-immunized mice [4]. From the results shown in Fig. 2 it could be concluded that most of the pepsin could be eliminated by passage of the collagen on a DEAE-Sepharose column. However, small amounts of pepsin present in the preparations could give rise to a T cell response as shown in Fig. 3. Immunization with lathyritic rat CII gave no anti-pepsin response (Fig. 3A) while immuniziation with pepsin-digested mouse
Eur. J. Immunol. 1990.20: 1061-1066 30
v p e I1 collagen-reactiveTcells in the mouse
1063
1
r
0 0
10
1
100
(B)
0.5
5
10
1
0 0
100
wlml
50
wml
Figure 1. Proliferative response of LN cells 7 days after immunization with 100 pg lathyritic rat CII in the hind footpads. Antigens were in (A) rat CII ( 0 )and mouse CII (m) and in (B) CNBr fragments CB8 (0),CBlO (0) and CBll (A) derived from chick CII.
Figure 3. Proliferative response of LN cells 7 days after immunization with 100 pg lathyritic rat CII (A) and 100 pg pepsin-digested mouse CII (B). Antigens were lathyritic rat CII (0),pepsindigested mouse CII).( and pepsin (0).
1201
0
2
10
50
250
Pglml 801
Figure 4. Proliferative response of LN cells 12 days after immunization with 200 pg lathyritic rat CII. Antigens were rat CII (0)and mouse CII (m).
60-
7
s x
40-
E
8
20
o
104 10-3 10-2 1 0 ’ ~ 1
10
100
Pglml
Figure2. Proliferative response of the T cell clone DCR34. Antigens were in (A) lathyritic rat CII (U), pepsin-digestedrat CII not passed on DEAE-Sepharose (0),pepsin-digested rat CII passed on DEAE-Sepharose (m) and in (B) pepsin (A).
whether this response was directed against a mouse CIIspecific epitope or against pepsin. However, using a very high antigen concentration a response against mouse CII could be detected also after immunization with pepsin-free rat CII. In Fig. 4 the response 12 days after immunization with 200 pg lathyritic rat CII is shown. At the highest concentration of CII (250 pg/ml) a proliferative response against mouse CII appeared.
3.2 Analysis of T cell lines CII gave a significant anti-pepsin response (Fig. 3B). To ascertain that pepsin-reactive T cells were not present among the proliferating cells it was therefore necessary to immunize mice with lathyritic rat CII prepared without pepsin digestion. Since we could not prepare lathyritic mouse CII the response after immunization with mouse CII could not be clearly analyzed. As shown in Fig. 3B there was no response against lathyritic rat CII after immunization with mouse CII while there was a significant response against mouse CII. It was not possible to determine
Since the autoreactivity in the primary cultures was only detected with high concentrations of CII, LN cells from mice immunized with lathyritic rat CII were repeatedly stimulated with high concentrations (50-100 pg/ml) or low concentrations (1-5 pg/ml) of rat CII. The response after four such passages is shown in Fig. 5. Both these lines reacted primarily with rat CII with high sensitivity and a high stimulation index (SI, cpm with antigen/cpm of medium control), SI = 130 at 0.4 p g / d for the line cultured with high concentrations of rat CII. The line cultured with
Eur. J. Immunol. 1990. 20: 1061-1066
M. Andersson and R. Holmdahl
1064
2oo
? ? E
1
100
4 0
0
0.01
0.1
1
10
100
Clslml
(B)
301
' 0
0.0180.08 O 0.4
0
2 O 10
50
250 L
ouc% 0
0.4
2
10
50
250
w m l
Calm1
Figure5. Proliferative response of T cells derived from mice immunized with rat CII. After a primary culture of the LN cells with rat CII at 100 pg/ml (A, B) or 25 pglml (C, D) the cells were stimulated on four consecutive occasions with 50 pg/ml of rat CII (A, B) or 1-5 pglml of rat CII (C, D) and then assayed. Antigens were rat CII ( 0 )and mouse CII (W). In (B) and (D) is shown the response against mouse CII at a higher concentration range only.
6ol
4g20j01 & f
? ?
20-
X
;
10-
0
0
0.025
0.25
2.5
25
nglrnl
Figure 7. Proliferative response of the T cell line M1. Antigens tested were: in (A) denatured rat CII (0),denatured bovine CII (W), native rat CII (0),native bovine CII ( O ) , native chick CII (A), native human CII (A), native mouse CII (0)and rat collagen I (V) and in (B) CNBr fragments CB8 (0),CBlO (W), CBll (0)and CB12 ( 0 )derived from chick CII.
50
4-
?i
(A)
30
10 "
.
.
.
.
0 0.0180.08 0.4
.
.
2
10
I
50
I
250
3-
? Y
2-
ti
Wml
Figure6. Proliferative response of T cells derived from mice immunized with rat CII. After a primary culture of the L N cells with mouse CII at 100 pglml the cells were repeatedly stimulated with 50 pglml of mouse CII. The assay was performed after six cycles of antigen stimulations. Antigens were rat CII (0) and mouse CII (W).
5-
(8)
4-
?
z
3-
Y
E, 2 0
high concentrations of CII still showed autoreactivity (SI = 6 at 50 pg/ml) while autoreactivity was not visible in the line cultured with low concentrations of CII. On the other hand, the same LN cells from mice immunized with lathyritic rat CII when repeatedly stimulated with 50-100 pg/ml mouse CII yielded a specific response primarily against mouse CII as shown in Fig. 6. The concentration needed and the SI were similar to those found for the autoreactivity when culturing the cells with high concentrations of rat CII (SI = 5 at 50 pg/ml). In this case the reactivity against rat CII was weaker than the autoreactive response. Several T cell lines were established from mice immunized with rat CII. None of them showed any autoreactivity. These lines were cultured with rat CII concentrations giving optimal proliferation (1-5 pg/ml).TheT cell line M1 which
1-
0-
I
was established in this manner was used for more thorough specificity analysis. After more than five cycles of expansion of T cells inT medium followed by antigen stimulation with rat CII and APC the proliferative response was analyzed. The results are shown in Fig. 7. The Tcell line
Eur. J. Immunol. 1990.20: 1061-1066
3001
/
q p e I1 collagen-reactive T cells in the mouse
1065
globulin gives a much stronger T cell response than autologous thyroglobulin [26]. It was also shown that a T cell line which did not proliferate in response to autologous thyroglobulin could induce thyroiditis upon transfer [27].
We have been working on the CIA model using the same approach. Earlier we presented data on CII-reactive Tcell lines capable of transferring disease to naive recipients [4]. An arthritogenic clone could not, however, be isolated. 0’ 0 1 10 100 After further characterizing the T cell response against CII figlml in DBA/l mice we have now shown that there is a Figure 9. Proliferative response in a secondary culture of SC from pronounced difference in the response against heterolonaive DBA/1 mice. The SC were cultured with mouse CII at a gous vs. autologous CII. Upon immunization with heteroconcentration of 50 pg/ml and then expanded in T medium. The logous CII, we could isolate T cells which proliferated in blasts were then assayed in a proliferation assay with rat CII (0) response to all heterologous CII tested at concentrations and mouse CII (m). < 0.1 pg/ml but not to mouse CII. The presence of autoreactive T cells could be demonstrated but these cells showed only a weak proliferation and only at high concenresponded strongly to rat, bovine, human and chick CII but trations of CII. We were able to show that the earlier Tcell not at all to mouse CII (Fig. 7A). Denatured CII gave a lines and clones described as autoreactive contained Tcells somewhat better response than native CII. No response was reacting with pepsin and we believe that this reactivity was seen with rat type I collagen. The proliferative response responsible for most of the reactivity against mouse CII. against different CNBr peptides derived from chick CII is However, even in the absence of proliferation against shown in Fig. 7B. The response was directed against the autologous CII autoreactiveTcells may be present in the T CBll peptide. cell lines as bystander cells only capable of proliferating if other cells producing the relevant growth factors are present. Such a possibility would be in accordance with the 3.3 T cell response of SC from naive mice in vitro model of Tcell tolerance described by Jenkins et al., where the tolerant T cell could be activated by its antigen As described by others [15] autoreactive T cells can be and express IL 2R but was incapable of producing IL 2 and detected in spleens of naive mice.We therefore analyzed SC therefore did not proliferate [28]. This might also account from naive DBA/1 mice for CII reactivity. The SC were for (a) the transfer of thyroiditis with aTcell line incapable cultured for 6 days, pulsed with [3H]dThd and harvested. of proliferating in response to autologous thyroglobulin The results of such an assay are shown in Fig. 8A. A [27] and (b) the fact that autologous MBP could trigger L N significant response against mouse CII and rat CII but not cells from MPB-primed rats to transfer encephalomyelitis rat type1 collagen was found. The response could be in the absence of a proliferative response [29]. Another inhibited by the addition of 10 pg/ml of the anti-CD4 mAb feature of the Jenkins model for clonal anergy which fits GK1.5 (Fig. 8B). To further verify the Tcell nature of the with our data is that the unresponsiveness to antigen can be response, SC cultured with 50 pg/ml of mouse CII were at least partly overcome with a high amount of antigen expanded in Tmedium and then assayed. As seen in Fig. 9 ~301. these cells showed an increased response against mouse CII and rat CII. The presence of two types of T cells may explain the difference in immunogenicity between heterologous and autologous CII. Immunization with heterologous CII gives rise to aTcell response primarily directed against a foreign 4 Discussion determinant on the CII molecule.TheseTcells interact with Self-nonself discrimination in the immune system is still a cross-reactive B cells [7] inducing a strong autoantibody matter of debate (reviewed in [20]). Both negative selection response. We were able t o show that this reactivity was of T cells based on self recognition in the thymus [21-231 as directed primarily against the C B l l fragment. This fragwell as peripheral tolerance in the absence of deletion [24] ment has earlier been shown to be the most immunogenic have been demonstrated. It would be reasonable to assume part of the CII molecule [31]. Immunization with autolothat these two mechanisms complement each other with gous CII on the other hand may trigger the autoreactive clonal deletion in the thymus acting on T cells recognizing anti-CII Tcell but since these cells show a weaker response ubiquitous antigens and peripheral tolerance acting on T the antibody response becomes much weaker [8,9]. cells recognizing more unevenly distributed antigens. If the T cells recognizing autologous CII in the DBA/l Autoreactive T cells have been shown t o be involved in mouse actually exhibit the type of clonal anergy discussed several experimental autoimmune disease models. The earlier this means that they have already interacted with most studied model in this respect is the experimental their antigen in the peripheral immune system of the naive autoimmune encephalomyelitis model. Autoreactive T cell mouse. This in turn suggests that it should actually be clones are readily isolated from rats and mice immunized possible t o detect them without previous immunization. In with myelin basic protein (MBP) and these T cells will a study of Hooper et al. [15] it was shown that Tcells from induce disaese [12] upon transfer to naive recipients. In the spleens of unprimed mice would proliferate in response experimental autoimmune thyroiditis similar data have to autologous erythrocytes. Using a similar protocol we been produced [25] but in this model heterologous thyro- were able to show proliferation against autologous CII in 1
1066
M. Andersson and R. Holmdahl
SC from DBA/1 mice. The response could be inhibited by the anti-CD4 antibody GK1.5. Although significant, the response against rat CII was not as strong as that against autologous CII. The proliferating cells were expanded in ILZcontaining medium and in a secondary culture an enrichment of autoreactiveTcells could be demonstrated. From the fact that in the unimmunized mouse the response was stronger against autologous CII than against heterologous CII we conclude that the autoreactive cells must have been activated in vivo. Based on the data presented here we would like to propose the followinghypothesis: autoreactive anti-CII T cells exist in the DBA/1 mouse which are not deleted but are present in the periphery. These T cells are subject to induction of peripheral tolerance and cannot be readily induced to proliferate. The fact that they can be detected as proliferating cells in cultures from spleens of unprimed mice indicate that activation is a step in the tolerance induction. However, the tolerance is not absolute and the Tcells can upon immunization be induced to perform certain effector functions, such as B cell help and DTH reactions, and thereby participate in the induction of arthritis. Immunization with heterologous CII will also stimulate T cells which are not autoreactive and therefore not regulated. Activation of these T cells might further enhance the activation of the autoreactive T cells. Further analysis of these two types of Tcells will be the subject of our future work. Received November 30, 1989; in revised form January 15, 1990.
5 References 1 Courtenay, J. S., Dallman, M. J., Dayan, A. D., Martin, A. and Mosedale, B., Nature 1980. 283: 666. 2 Wooley, F! H., Luthra, H. S., Stuart, J. M. and David, C. S., J. Exp. Med. 1981. 154: 688. 3 Ranges, G. E., Sriram, S. and Cooper, S. M., J. Exp. Med. 1985.162: 1105. 4 Holmdahl, R., Klareskog, L., Rubin, K., Larsson, E. and Wigzell, H., Scand. J. Immunol. 1985. 22: 295. 5 Holmdahl, R., Jonsson, R., Larsson, F! and Klareskog, L., Lab. Invest. 1988. 58: 53. 6 Stuart, J. M. and Dixon, F. J., J. Exp. Med. 1983.158: 378.
Eur. J. Immunol. 1990. 20: 1061-1066
7 Holmdahl, R., Jansson, L., Gullberg, D., Rubin, K., Forsberg, l? 0. and Klareskog, L., Clin. Exp. Immunol. 1985. 62: 639. 8 Holmdahl, R., Klareskog, L., Andersson, M. and Hansen, C, Immunogenetics 1986. 24: 84. 9 Holmdahl, R., Jansson, L., Andersson, M. and Larsson, E., Immunology 1988. 65: 305. 10 Holmdahl, R., Jansson, L., Larsson, E., Rubin, K. and Klareskog, L., Arthritis Rheum. 1986. 29: 106. Carlioz, A., Roudier, R. and 11 Boissier, M.-C., Feng, X.-Z., Fournier, C., Ann. Rheum. Dis. 1987. 46: 691. 12 Ben-Nun, A.,Wekerle, H. and Cohen, I. R., Eur. J. lmmunol. 1981. 11: 195. 13 Holmdahl R., Andersson, M., Enander, I., Goldschmidt,T., Jansson, L., Larsson, F!, Mo, J., Nordling, C. and Klareskog, L., Int. Rev. Immunol. 1988. 4: 49. 14 Coutinho, A., Marquez, C., Araujo, F? M. F., Pereira, I?, Toribio, M. L., Marcos, M. A. R. andMartinez-A., C., Eur. J. Immunol. 1987. 17: 821. 15 Hooper, D. G. and Taylor, R. B., Eur. J. Immunol. 1987. 17: 797. 16 Smith, B. D., Martin, G. R., Miller, E. J., Dorfman, A. and Swarm, R., Arch. Biochem. Biophys. 1975.166: 181. 17 Miller, E. J. and Rhodes, R. K., Methoak Enzymol. 1982.82: 33. 18 Terato, K., Cremer, M. A., Hasty, K. A., Kang, A. H., Hasty, D. L. and Townes, A. S., Collagen Rel. Res. 1985. 5: 469. 19 Dialynas, D. F! ,Wilde, D. B., Marrack, F!,Pierres, A., Wall, K. A., Havran,W. L., Otten, G. R., Loken, M. R., Pierres, M., Kappler, J. W. and Fitch, E W., J. lmmunol. 1983. 131: 2445. 20 Schwartz, R. H., Cell 1989. 57: 1073. 21 Kappler, J. W., Roehm, N. and Marrack, F!, Cell 1987. 49: 273. 22 MacDonald, H R., Schneider, R., Lees, R. K., Howe, R. C., Acha-Orbea, H., Festenstein, H., Zinkernagel, R. M. and Hengartner, H., Nature 1988. 332: 40. 23 Kisielow, F!, Bliithmann, H., Staerz, U. D., Steinmetz, M. and Von Boehmer, H., Nature 1988. 333: 742. 24 Burkly, L. C., Lo, D., Kanagawa, O., Brinster, R. L. and Flavell, R. A., Nature 1989.342: 564. 25 Romball, C. G. and Weigle, W. O., J. Immunol. 1987. 138: 1092. 26 Romball, C. G. and Weigle,W. O., Eur. J. Immunol. 1984.14: 887. 27 Maron, R., Zerubavel, R., Friedman, A. and Cohen, I. R., J. Immunol. 1983. 131: 2316. 28 Jenkins, M. K. and Schwartz, R. H., J. Exp. Med. 1987. 165: 302. 29 Hunter, S., Cell. Immunol. 1986. 97: 204. 30 Jenkins, M. K., Ashwell, J. D. and Schwartz, R. H., J. lmmunol. 1988.140: 3324. 31 Terato, K., Hasty, K. A., Cremer, M. A., Stuart, J. M.,Townes, A. S. and Kang, A. H., J. Exp. Med. 1985. 162: 637.
