spontaneous autoimmunity in transgenic mice

Proc. Nati. Acad. Sci. USA Vol. 89, pp. 2985-2989, April 1992 Immunology

T-cell responsiveness to an oncogenic peripheral protein and spontaneous autoimmunity in transgenic mice (tolerance/simian virus 40)

TERRENCE GEIGER*, LINDA R. GOODINGt, AND RICHARD A. FLAVELL*t tHoward Hughes Medical Institute and *Section of Immunobiology, Yale University School of Medicine, 310 Cedar Street, New Haven, CT 06510; and

tDepartment of Microbiology and Immunology,

Emory University School of Medicine, Atlanta, GA 30322

Communicated by Vincent T. Marchesi, December 9, 1991

ABSTRACT Why T cells develop autoimmune reactivity to some antigens and tolerance to others is unknown. Various mechanisms can provide for T-cell tolerance. These include deletion in the thymus, exhaustive differentiation in the periphery, T-cell receptor and coreceptor downregulation, and anergy. Which mechanisms normally provide for tolerance to antigens expressed on specific tissues and why they sometimes fail is unclear. To understand this, we analyzed how a tissuespecific protein with defined timing and location of expression is recognized by T cells so as to induce tolerance or autoimmunity. We crossed mice expressing the simian virus 40 large tumor antigen on pancreatic acini beginning 4-25 days after birth with mice transgenic for a rearranged T-cell receptor that recognizes this antigen presented by the class I major histocompatibility complex molecule H-2Kk. No T-cell tolerance was found; rather, T-cell reactivity accompanied lymphocytic infiltration and pancreatic acinar destruction. This result argues that T cells may become spontaneously autoreactive to certain postnatally expressed peripheral proteins and that this reactivity may lead to autoimmune disease.

tolerance in adults to antigen, by immunizing within certain dose ranges, or to grafts, by removing passenger leukocytes, further showed that antigen encounter in adults is not inherently immunogenic (13-16). These results suggested that how, rather than when, antigen is encountered distinguishes the induction of tolerance from the induction of immunity. We describe here transgenic mice expressing the simian virus 40 large tumor antigen (SV-T) on pancreatic acini beginning 4-25 days after birth and a transgenic rearranged T-cell receptor (TCR), which recognizes this antigen presented by the class I MHC H-2Kk. We observe T-cell reactivity to antigen in vitro as well as pancreatic infiltration and destruction. This phenotype differs from all other transgenic models of T-cell tolerance studied to date and affirms the importance of the timing of an antigen's expression in distinguishing tolerance from autoimmunity.

MATERIALS AND METHODS TCR Cloning. RNA was prepared from the cytoxic T lymphocyte (CTL) clone K as described (17). p-chain cDNA was then prepared by PCR using primers specific for the V,98.1 variable (V) region and the p-chain constant (C) region. a-chain cDNA was prepared using anchorage-dependent PCR as described (18). DNA sequencing showed the a-chain V region to be VGt5H and confirmed the p chain as V,38. 1. The construction of synthetic genomic TCR constructs is de-

The T-cell response to tissue-specific antigens is poorly understood. By using tissue-specific promoters such as the insulin or the elastase promoters to localize antigens, transgenic mouse studies have been particularly useful in documenting the T-cell response to these antigens. Virtually all these models have studied T-cell tolerance to allogeneic major histocompatibility complex (MHC) molecules. Some tissue-specific MHC molecules induce T-cell tolerance (1-4). Others are "ignored" by T cells (5-7). In the latter case, T cells capable of responding in vitro or upon immunization do not respond to the tissue-specific antigen in vivo. The nonMHC glycoprotein or nucleoprotein from lymphocytic choriomeningitis virus expressed in pancreatic islets with the rat insulin promoter also results in T-cell ignorance (8, 9). It is unclear which features of tissue-specific antigens influence the pattern of T-cell tolerance observed. We were interested in the impact of an antigen's developmental timing of expression on tolerance. Grafting studies, in which recipients were primed with donor cells before grafting, indicated that antigen first administered prenatally or neonatally induced tolerance, whereas that first administered later induced immunity (1012). For the male-specific H-Y antigen, the ability to induce tolerance by priming with H-Y-expressing cells dropped off between postnatal days 5 and 22. This work established the notion that prenatal or neonatal antigen encounter is critical in the establishment of tolerance. Later work, however, disputed this notion. The ability to induce specific immunity in fetuses showed that antigens first encountered prenatally are not inherently tolerizing (13). The ability to induce

scribed in the text. Proliferation Assay. Either 3 x 104 cells or the designated number of lymph node responder cells were added to 104 or the designated number of stimulator cells and 5 x 105 syngeneic, 2000-rad irradiated (1 rad = 0.01 Gy) splenocytes in Click's medium (Irvine Scientific) with 5% fetal calf serum using flat-bottom 96-well plates. Stimulators used were either 5000-rad irradiated PS-C3H cells, a simian virus 40transformed cell line, or 8000-rad irradiated non-simian virus 40-transformed syngeneic Ltk- cells. Proliferation was measured after a 3-day stimulation by [3H]thymidine incorporation. All samples were assayed in duplicate. Cytolysis Assay. As targets, PS-C3H or Ltk- cells were harvested and labeled for 2 hr at room temperature in Na51Cr/saline. These cells were washed three times before use. As effectors, 4 x 10' splenocytes were cocultured with 105 PS-C3H cells in a well of a 24-well plate. Four days after stimulation, effector cells were harvested, counted, and added to 104 target cells in round-bottom 96-well plates to produce the designated effector/target cell ratio. Assays were performed in Click's medium (Irvine Scientific) with 5% fetal calf serum. In maximum release wells, 104 target cells in 0.1 ml were added to 0.1 ml of 0.1 M HCL. In spontaneous release wells, 104 target cells in 0.1 ml were added to 0.1 ml

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Abbreviations: MHC, major histocompatibility complex; SV-T, simian virus 40 large tumor antigen; TCR, T-cell receptor; V, variable; C, constant; J, joining; CTL, cytotoxic T lymphocyte.

2985

Immunology: Geiger et al.

2986

A ,i i

Sail

Proc. Natl. Acad. Sci. USA 89 (1992)

B

~ LVJ

Bglll

Pstl

EcoRi

FRAGMENT Sall - BgIll

Xmal

Xhol

Sad

EcoRl

FRAGMENT Sacl -EcoR

BaVDJ Kpn

BgIlI

Styl

2C

SORFCE

Bgil

StyL

A

HindIll

SIZE (bp) 3200 f

it-7

f

Kpni

23 0

2C

EcoRi - Styl

300

2C

Pstl - EcoRi

26 0

OUGO

Styl - BgIll

200

OUGO

EcoRI

300

PCR,KcDNA

BgIll - Kpnl

300

Hindell - Xmal

60

OUGO

Kpnl - BamHI

100-

OUGO

Xmal - Xhol

2000

PCR, 2C

BamHI

300

PCR,2C

HindIll

-

Sacll

Sacll

SIZE (bp) SOURCE 2800 72

BgIll - Pstl

-

BamHI

SAL

PCR,KcDNA

FIG. 1. Synthetic genomic TCR constructs. The replacement constructs containing clone K VJ and VDJ sequence (D, diversity) and 2C flanking sequence are shown. They were produced as described in the text. (A) a-chain replacement construct. (B) /3-chain replacement construct. *, Site created by mutagenesis; bp, base pairs; OLIGO, synthetic oligonucleotide.

of medium. After a 5-hr incubation at 370C, 0.1-ml aliquots were harvested and counted. Specific cytotoxicity was calculated as 100 x (sample cpm - spontaneous cpm)/ (maximum cpm - spontaneous cpm). Flow Cytometry. Three-color staining was performed on thymocytes and lymph node cells using CD8-fluorescein isothiocyanate (FITC) (Becton Dickinson), CD4-red613 (GIBCO/BRL), and KJ16(V#8.1 + -8.2) biotin in phosphatebuffered saline with 2% fetal calf serum, followed by streptavidin-phycoerythrin (Southern Biotechnology Associates, Birmingham, AL). Haplotyping was performed by staining thymocytes with Y-25 antibody (H-2Kb specific) followed by goat antimouse IgG-FITC or antiH-2Kk-FITC (Becton Dickinson). Samples were analyzed on a Becton Dickinson FACscan.

region cosmid C41/45 (21). These synthetic a and /3 constructs were coinjected into (C57BL/6J x CBA/J)F2 day-1 embryos to produce transgenic mice. Mice were subse-

A

.s

I

RESULTS Production and Characterization of TCR Transgenic Mice. An analysis of T-cell tolerance benefits from a sufficiently high T-cell precursor frequency to analyze antigen-reactive cells by using T-cell-specific antibodies. To obtain this, we made TCR transgenic mice containing both rearranged V,^ and Vp genes from a CTL clone that recognizes SV-T. Genomic DNA clones of the rearranged TCR from the V,98.1', SV-T-reactive, H-2Kk-restricted CTL clone K (19, 20) was prepared by a synthetic approach. This synthesis replaced the V and J (joining) regions from a construct successfully used to produce TCR transgenic mice, that of the 2C TCR (21), with the V and J regions of the clone K TCR. This was done as follows: PCR mutagenesis was used to introduce restriction sites into the clone K a and ,f cDNAs such that the TCR amino acid sequence was unaltered. Similarly, restriction sites were placed in fragments from genomic 2C TCR a- and p-chain constructs (21), both upstream and downstream of the VJ sequences. Synthetic oligonucleotides were used to bridge the gap in nucleic acid sequence between the clone K cDNA sequence and that of the 2C TCR upstream and downstream regions. These oligonucleotides re-created the Va and Va introns, coding sequence of the V and J regions, functional splice sites, and 5' promoter and 3' intronic sequence until linkage with 2C-derived sequence. The composition of these "replacement constructs" is shown in Fig. 1. The a-chain replacement construct was then ligated into the a-chain C region cosmid CGBS2. The /8-chain replacement construct was ligated to the plasmid p4B6 to provide 11.2 kilobases of additional 5' flanking sequence and then to the 83-chain C

B

-.4 FIG. 2. Flow cytometric analysis of lymph node cells (A) and thymocytes (B) from 10-week-old nontransgenic, TCR transgenic, and TCR/SV-T double transgenic littermates. Three-color staining was performed with CD8, CD4, and Vp8.1 + -8.2 (KJ16)-specific antibodies. The TCR and TCR/SV-T mice were MHC typed as H-2Kk/k, and the nontransgenic mouse was typed as H-2Kk/b. Results equivalent to those shown here were found in three other sets of single and double transgenic mice.



quently bred with B1O.BR and progeny were analyzed by Southern blotting. Three transgenic lines were obtained, one of which, named TG-B, was further characterized. The T-cell repertoire of the TG-B lineage is predominantly transgenic. Three-color flow cytometry with CD4-, CD8-, and V,8-specific antibodies showed that the vast majority of lymph node and thymic T cells express V,8 (93% peripherally in Fig. 2A and 96% thymically in Fig. 2B, numbers that are typical of the TG-B lineage). The surface TCR level, as determined by staining intensity, is equivalent in TG-B and nontransgenic mice. As has been shown in previously characterized class I-restricted TCR transgenic mice, the CD4/ CD8 ratio of peripheral TG-B T cells is heavily skewed toward CD8' cells (1:22 in transgenic versus 1.5:1 in control; Fig. 2A) (21-23). This suggests that virtually all the cells also express the transgenic Va, with cells Vp8+CD8' expressing both the Va, and the Vp transgenes being directed into the CD8' lineage by positive selection. We have eliminated a second possibility that the transgenic independent of the transgenic Va, directs cells into the Vp, CD8' lineage, causing the skewing of the CD4/CD8 ratio by the following experiment: We bred TG-B transgenic mice with mice of the aT2 lineage (24), which expresses SV-T using the crystallin promoter. We have shown by PCR that aT2 mice express SV-T on cells of hematopoietic origin (data not shown). Double-transgenic TCR/aT2 mice consequentially thymically delete SV-T-reactive cells; thymus size is reduced from 5 x 107 thymocytes in a TCR transgenic mouse to 106 thymocytes in a double transgenic mouse. The residual mature T cells in a TCR/aT2 mouse primarily express V,8 but have a CD4/CD8 ratio of 1.8:1. Thus, the presence of the transgenic Vp is not in itself sufficient to induce a skewed CD4/CD8 ratio. TG-B T cells functionally recognize SV-T. Unprimed peripheral lymph node cells from transgenic and nontransgenic mice were stimulated with the H-2k, SV-T-expressing PSC3H cell line, or the syngeneic, non-SV-T-expressing Ltkcell line. While unprimed transgenic cells proliferate in response to the PS-C3H cells, nontransgenic cells do not. No response is seen to the Ltk- cells by lymph node cells from either mouse (Fig. 3). Likewise, in a cytolysis assay, trans106 1

2987

-V- NON-TRANSGENIC (PS-C3H) TCR (PS-C3H) ..Z,-0- TCR/SV-T (PS-C3H) x 25--* NON-TRANSGENIC (Ltk) 35

TCRISV-T (Ltk)

15-

o)

0-0 5*5 -513 40

1.5

4.5

E/T ratio FIG. 4. Cytolytic response of transgenic splenocytes against SV-T-expressing cells. Splenocyte effector cells were cocultured with SV-T-expressing PS-C3H cells before analysis of cytolytic activity. Comparable levels of killing of PS-C3H cells are observed by TCR and TCR/SV-T transgenic mice. No killing of Ltk- cells, which do not express SV-T, is seen. Specific cytotoxicity is plotted on the ordinate. The ratio of effector/target (E/T) cells is plotted on the abscissa. The relatively low level of killing shown here is consistently found with TCR transgenic cells. Other SV-Texpressing cells or Ltk- cells incubated with antigenic peptide (19) do not serve as better targets (S. Guerder, personal communication).

genic but not nontransgenic cells are capable of killing the PS-C3H cell line (Fig. 4). Autoimmunity in TCR/SV-T Double Transgenic Mice. TCR transgenic mice were bred with transgenic mice expressing SV-T on pancreatic acini under the control of the pancreatic elastase promoter. This SV-T transgenic line, Tg(Ela1,SV40E)BRI18 (177-5) (25), expresses SV-T beginning 4-25 days after birth. T-cell response was analyzed in 10- to 12-week-old progeny. At this time, there is prominent SVT-induced dysplasia but no pancreatic tumors in the SV-Texpressing mice (Fig. 5). Thymocytes from TCR/SV-T double transgenic mice were analyzed by three-color flow cytometry using CD4-, CD8-, and V,98-specific antibodies. Flow cytometry profiles are

A

106I

I

B

1 05.

-

NON TRANSGENIC

*

SV-T

0

TCR

+

TCR/SV-T

105"

104

0

E 0

-0--O

1 o2

NON TRANSGENIC (PS-C3H) TCR (PS-C3H) TCR / SV-T (PS-C3H) NON TRANSGENIC (Ltk) TCR (Ltk) TCR / SV-T (Ltk)

=l~p

101 -

103

104 Lymph node responders

105 per

well

20

40

60

80

1(

120

PS-CSH stimulators

FIG. 3. Proliferative response of lymph node cells to SV-T. The proliferative response of lymph node cells from nontransgenic, TCR transgenic, SV-T transgenic, and TCR/SV-T transgenic mice to the SV-T transformed cell line PS-C3H or the nontransformed syngeneic cell line Ltk- was measured. (A) Titration of lymph node responders shows equivalent response by TCR and TCR/SV-T transgenic mice to PS-C3H cells. TCR and TCR/SV-T mice were H-2k/k, and the nontransgenic mouse was H-2k/b. Mice used were littermates. The high baseline incorporation in PS-C3H-containing wells relative to Ltk- containing wells is seen in the absence of responder cells (data not shown) and therefore reflects background thymidine uptake. This is likely due to the irradiated PS-C3H cells, and not the irradiated Ltk- cells, detaching from the plates before harvesting. Both cell lines are normally adherent. (B) Titration of PS-C3H stimulator cells in the presence of 3 x 10" lymph node responders. Abscissa indicates number, in thousands, of PS-C3H stimulator cells. Littermates, all of which were typed as H-2Kk/b, were used. The addition of 10 units of recombinant human interleukin 2 per ml had no effect on relative proliferation between the TCR and TCR/SV-T mice (data not shown).

2988


similar and within the limits of variability seen with TCR transgenic mice alone (Fig. 2B). Antigen reactivity among thymocytes, as determined by proliferation to the SV-Texpressing PS-C3H cell line, is also equivalent in TCR and TCR/SV-T mice (data not shown). Thus, thymocyte development is unaffected by the pancreatic SV-T. Flow cytometry was then performed on peripheral lymphocytes. TCR and TCR/SV-T lymph node cells appear to be similar (Fig. 2A). There is no reduction in CD8 or TCR levels on the peripheral T cells and the V,98'CD8' population


remains predominant. The slight decrease in V,98'CD8' cells as a fraction of the total lymph node population seen in the

TCR/SV-T mouse compared with the TCR littermate is also seen in other 10- to 12-week-old mice. This may reflect differences in T-cell circulation patterns between the TCR and TCR/SV-T transgenic mice or peripheral deletion of TCR/SV-T antigen-reactive T cells upon antigen stimulation (26). Stimulation of lymph node responder cells from TCR/ SV-T or TCR mice with the SV-T-expressing cell line PSC3H demonstrated a similar response to antigen in both mice (Fig. 3). Titrations of both antigen amount and responder cell numbers reveals similar dose-response relationships. Neither mouse responds to the non-SV-T-expressing syngeneic Ltk- cell line. Killing assays using 51Cr-labeled PS-C3H or Ltk- cells as targets further demonstrated an equivalent functional response in TCR and TCR/SV-T mice (Fig. 4). Thus, unlike multiple examples of T-cell tolerance to peripherally expressed MHC molecules (1, 4, 27), TCR/SV-T mice show no apparent influence of the peripheral antigen on T-cell phenotype or function. To determine whether we were observing autoimmunity or the T-cell ignorance seen by others, in which the pancreas would be unscathed despite the in vitro T-cell responsiveness (5-8), transgenic pancreases were formalin fixed and stained with hematoxylin and eosin. The double transgenic pancreas shown has a diffuse immune infiltrate and concomitant exocrine tissue destruction (Fig. 5C). Thus, the absence of in vitro T-cell tolerance here accompanies autoimmunity. Of seven other 10- to 20-week-old double transgenic pancreases examined, three showed similar features-one showed complete exocrine pancreas destruction with no visible exocrine tissue, two showed parenchymal infiltrate with little exocrine tissue destruction, and one showed primarily a perivascular infiltrate. Response in SV-T Single Transgenic Mice. Analyses were also performed on SV-T single transgenic animals to determine whether they develop autoimmune manifestations similar to those of the TCR/SV-T mice. Peripheral lymphocytes from these animals were found to be unresponsive to antigen in vitro (Fig. 3B). This suggests that there is no in vivo priming of SV-T-reactive cells or that such cells are inactivated or present in low numbers. Without a marker for antigenreactive T cells, distinguishing between these possibilities is difficult. Sectioning of SV-T transgenic pancreases revealed an infiltrate primarily localized to perivascular regions (Fig. 5B). Immunization with SV-T-expressing PS-C3H cells did not induce an infiltrate more extensive than that seen in unimmunized mice (data not shown). Thus, a low-grade inflammation is occurring in the SV-T mice and the presence of transgenic antigen-reactive T cells may be amplifying this in the TCR/SV-T mice. The mechanism of such autoimmune amplification may involve an overloading of elements restraining immune activation by the anomalously high precursor frequency of SV-T-reactive cells.

DISCUSSION To understand how tissue-specific antigen may induce either FIG. 5. Hematoxylin and eosin staining of pancreatic sections. (A) Staining of a formalin-fixed section from a TCR transgenic mouse displays normal pancreatic architecture without inflammation. (B) A region of a 10-week-old SV-T mouse pancreas showing a focal perivascular infiltrate. Atypia in cell nuclei and acinar architectural distortion is also apparent. (C) Representative field from a 10-weekold TCR/SV-T mouse is shown. The pancreas appears edematous and exocrine tissue destruction with a diffuse infiltrate is seen.

T-cell tolerance or autoimmunity, we crossed mice transgenic for a rearranged TCR recognizing the SV-T antigen with mice expressing SV-T on pancreatic acini beginning 4-25 days after birth. We have shown that T cells in such double transgenic mice remain fully responsive to antigen. T cells were both capable of proliferating to and lysing SV-Texpressing cells. Furthermore, pancreatic destruction is observed. This result differs from the tolerance observed to class I or class II MHC molecules expressed in a tissue-specific man-

Immunology: Geiger et al. ner. The MHC class II I-E, expressed on p islet cells using the insulin promoter, was shown to generate tolerance by the induction of anergy (27). The class I H-2Kb molecule, when expressed on the same tissue, induced a state of cytokine dependence-T cells responding to antigen only in the presence of exogenous interleukin 2 (4). The class I H-2Kb molecule when expressed in the brain with the glial fibrillary acidic protein promoter provoked T-cell tolerance by downregulating the TCR and coreceptor molecules (1). Our results also differ from those obtained when other class II MHC molecules or the lymphocytic choriomeningitis virus glycoproteins or nucleoproteins were expressed on 13 islet cells (5, 6, 8, 9). In these cases, unprimed T cells were antigen reactive in vitro, as shown by proliferation and/or cytolysis assays, but did not respond to the tissue-specific antigen in vivo. Thus, these T cells either lacked the ability to access antigen, or antigen encounter was a neutral event, leading neither to tolerance nor to activation. This study supports the hypothesis that the T-cell response to tissue-specific antigens may follow three courses (28). If the antigen is prenatally expressed, T cells will become tolerant and will not respond to antigen in vivo or in vitro. We observe this unresponsiveness in T cells of the RipTag2 transgenic mouse lineage, which express SV-T on P islet cells fetally (unpublished data). If the antigen is expressed postnatally, T cells will remain reactive in vitro. The in vivo response, however, will depend on an appropriate priming environment. The dysplasia and cell death observed in the SV-T pancreas here may provide a milieu in which an autoimmune response is primed (28). Local cytokine production due to nonspecific inflammation may provide the necessary activation signals to prime this response (29). Spontaneous autoimmunity has been seen in other tissuespecific transgenic models and in one case this was correlated with antigen expression beginning 10-12 weeks after birth (30, 31). In these models, an autoantibody response accompanied tissue infiltration and destruction. T-cell responses were not analyzed. Importantly, because T-cell tolerance can coexist with B-cell responsiveness, the status of the T-cell compartment cannot be implied from the autoantibody responses (32). Current models of T-cell tolerance suggest that tolerance to tissue-specific proteins is due to the signaling properties of the antigen presenting cell-the intrinsic inability of most peripheral tissues to provide certain secondary signals inducing T-cell tolerance (33). Our results suggest that in addition, the developmental timing of antigen expression is critical in distinguishing the development of tolerance from autoimmunity. Potentially there exist pathways of antigen processing or intercellular transfer of antigen that operate fetally and promote the establishment of tolerance even among T cells first released into the periphery postnatally. Antigens first expressed postnatally, as SV-T here, would be excluded from these pathways. We hope this transgenic approach will help simplify the analysis ofT-cell involvement in autoimmunity. We give special thanks to Charlie Janeway, Rob Homer, Sylvie Guerder, and Nancy Maizels for helpful advice and critical manuscript review; to Frances Rawle for excellent teaching; to Cindy Hughes and Debbie Butkus for DNA injection and embryo transfers; to Bill Sha and Dennis Loh for providing the 2C TCR cosmids and subclones; to Ralph Brinster for providing the Tg(Ela-1,SV40E)BRI18 mice; and to Heiner Westphal for providing the aT2 transgenic mice. Histology sections were made at the Yale University Department of Pathology. T.G. was supported by the Medical Scientist Training Program at Yale University and by a fellowship


2989

from Miles Pharmaceuticals. R.A.F. is an investigator for the Howard Hughes Medical Institute 1. Schonrich, G., Kalinke, U., Momburg, F., Malissen, M., Hammerling, G. J. & Arnold, B. (1991) Cell 65, 293-304. 2. Lo, D., Burkly, L. C., Widera, G., Cowing, C., Flavell, R. A., Palmiter, R. D. & Brinster, R. L. (1988) Cell 53, 159-168. 3. Lo, D., Burkly, L. C., Flavell, R. A., Palmiter, R. D. & Brinster, R. L. (1989) J. Exp. Med. 170, 87-104. 4. Morahan, G., Allison, J. & Miller, J. F. A. P. (1989) Nature (London) 339, 622-624. 5. Bohme, J., Haskins, K., Stecha, P., van Ewjik, W., LeMeur, M., Gerlinger, P., Benoist, C. & Mathis, D. (1989) Science 244, 1179-1183. 6. Murphy, K. M., Weaver, C. T., Elish, M., Allen, P. M. & Loh, D. Y. (1989) Proc. Natl. Acad. Sci. USA 86, 10034-10038. 7. Morahan, G., Brennan, F. E., Bhathal, P., Allison, J., Cox, K. 0. & Miller, J. F. A. P. (1989) Proc. Natl. Acad. Sci. USA 86, 3782-3786. 8. Ohashi, P. S., Oehen, S., Buerki, K., Pircher, H., Ohashi, C. T., Odermatt, B., Malissen, B., Zinkernagel, R. M. & Hengartner, H. (1991) Cell 65, 305-317. 9. Oldstone, M. B. A., Nerenberg, M., Southern, P., Price, J. & Lewicki, H. (1991) Cell 65, 319-331. 10. Billingham, R. E., Brent, L. & Medawar, P. B. (1953) Nature (London) 172, 603-606. 11. Billingham, R. E. & Silvers, W. K. (1960) J. Immunol. 85, 14-26. 12. Medawar, P. B. & Woodruf, M. F. A. (1958) Immunology 1, 27-35. 13. Dresser, D. W. & Mitchison, N. A. (1968) Adv. Immunol. 8, 169-181. 14. Fink, P. J., Shimonkevtz, R. P. & Bevan, M. J. (1988) Annu. Rev. Immunol. 6, 115-137. 15. Lafferty, K. J., Prowse, S. J. & Simeonovic, C. J. (1983) Annu. Rev. Immunol. 1, 143-173. 16. Markmann, J., Lo, D., Naji, A., Palmiter, R. D., Brinster, R. L. & Heber-Katz, E. (1988) Nature (London) 336, 476-479. 17. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A. & Struhl, K., eds. (1989) Current Protocols in Molecular Biology (Wiley, New York). 18. Loh, E. Y., Elliott, J. F., Cwirla, S., Lanier, L. L. & Davis, M. M. (1989) Science 243, 217-220. 19. Rawle, F. C., O'Connell, K. A., Geib, R. W., Roberts, B. & Gooding, L. R. (1988) J. Immunol. 141, 2734-2739. 20. O'Connell, K. & Gooding, L. (1984) J. Immunol. 132, 953-958. 21. Sha, W. C., Nelson, C. A., Newberry, R. D., Kranz, D. M., Russell, J. H. & Loh, D. Y. (1988) Nature (London) 335, 271-274. 22. Teh, H. S., Kisielow, P., Scott, B., Kishi, H., Uematsu, Y., Bluthman, H. & von Boehmer, H. (1988) Nature (London) 335, 229-233. 23. Pircher, H., Burki, K., Lang, R., Hengartner, H. & Zinkernagel, R. (1989) Nature (London) 342, 559-561. 24. Mahon, K. A., Chepelinsky, A. B., Khillan, P. S., Overbeek, P. A., Piatigorsky, J. & Westphal, H. (1987) Science 235, 1622-1628. 25. Ornitz, D. M., Palmiter, R. D., Messing, A., Hammer, R. E., Pinkert, C. A. & Brinster, R. L. (1985) Cold Spring Harbor Symp. Quant. Biol. 50, 399-409. 26. Webb, S., Morris, C. & Sprent, J. (1990) Cell 63, 1249-1256. 27. Burkly, L. C., Lo, D., Kanagawa, O., Brinster, R. L. & Flavell, R. A. (1989) Nature (London) 342, 564-566. 28. Hanahan, D. (1990) Annu. Rev. Cell Biol. 6, 493-537. 29. Sarvetnick, N., Shizuru, J., Liggitt, D., Martin, L., McIntyre, B., Gregory, A., Parslow, T. & Stewart, T. (1990) Nature (London) 346, 844-847. 30. Adams, T. E., Alpert, S. & Hanahan, D. (1987) Nature (London) 325, 223-228. 31. Roman, L. M., Simons, L. F., Hammer, R. E., Sambrook, J. F. & Gething, M.-J. H. (1990) Cell 61, 383-396. 32. Adelstein, S., Pritchard-Briscoe, H., Anderson, T. A., Crosbie, J., Gammon, G., Loblay, R. H., A., B. & Goodnow, C. C. (1991) Science 251, 1223-1228. 33. Schwartz, R. S. (1990) Science 248, 1349-1356.