A Diabetogenic Gene Prevents T Cells from Receiving Costimulatory Signals

A Diabetogenic Gene Prevents T Cells from Receiving Costimulatory Signals

Cellular Immunology 194, 90 –97 (1999) Article ID cimm.1999.1501, available online at http://www.idealibrary.com on A Diabetogenic Gene Prevents T Ce...

122KB Sizes 0 Downloads 30 Views

Cellular Immunology 194, 90 –97 (1999) Article ID cimm.1999.1501, available online at http://www.idealibrary.com on

A Diabetogenic Gene Prevents T Cells from Receiving Costimulatory Signals Jodene K. Moore,* Daniel P. Gold,† Stephen C. Dreskin,‡ Åke Lernmark,§ and Donald Bellgrau* *The Barbara Davis Center for Childhood Diabetes, Department of Immunology, and ‡Department of Medicine, University of Colorado Health Sciences Center, 4200 E. Ninth Avenue, Denver, Colorado 80262; †The Sidney Kimmel Cancer Center, 3099 Science Park Road, Suite 200, San Diego, California 92121; and §Robert H. Williams Laboratory, Department of Medicine, University of Washington School of Medicine, 1959 Pacific Avenue, Seattle, Washington 98195 Received November 2, 1998; accepted March 30, 1999

mellitus (1). Selective inbreeding led to the development of two BB sublines (2). One, termed diabetesprone BB (BB-DP), displays essentially complete disease penetrance. Animals develop diabetes by 120 days of life, with the proviso that the colony is maintained under specific pathogen-free conditions (3). Diabetesresistant BB rats (BB-DR) normally do not develop diabetes spontaneously; however, these animals can become diabetic following the presentation of inductive stimuli such as infection with Kilham’s rat virus (3– 6). The BB-DP and BB-DR sublines appear to differ in diabetogenicity in one diabetes-susceptibility gene, designated in the rat as iddm1 (7, 8). Iddm1 encodes for a genetic defect mapping to chromosome 4 (8) that results in peripheral T cell lymphopenia (9 –11). The existing peripheral T cells in the lymphopenic animal display a distinctive cell surface phenotype wherein virtually all T cells express low levels of the RT6 (12) and CD45RC molecules (13). This phenotype is unusual in that the majority of mature peripheral lymphocytes in nonlymphopenic rats, including the BB-DR strain, bear high levels of RT6 and CD45RC (14 –17). During rat T cell development, Thy1 is expressed on thymocytes and recent thymic emigrants (16, 17). Loss of Thy1 expression on peripheral T cells is associated with coordinate expression of RT6 and occurs within 7 days of emigration from the thymus. The absence of RT6 expression concurrent with an increased frequency of Thy1 1 T cells can be interpreted to indicate that a majority of peripheral T cells in the BB-DP rat are immature; phenotypically, these cells resemble the recent thymic emigrant stage of T cell development. Transfer of peripheral T cells from the BB-DR to the BB-DP can protect the BB-DP rats from disease (18). The protective effect of adoptive transfer of BB-DR T cells into BB-DP recipients has been attributed to the RT6 1 T cell subset, which is severely deficient in BB-DP rats, but normal in BB-DR rats. Current theory postulates that the RT6 1 subset contains cells that

T cell fate following antigen encounter is determined by several intracellular signals generated by the interaction of the T cell with an antigen-presenting cell. In the periphery activation requires T cell receptor signaling (signal one) in combination with costimulatory signals (signal two), usually provided through the cognate interaction of CD28 and B7 molecules. Provision of signal one alone to purified murine peripheral T cells in vitro induces apoptosis or anergy rather than promoting activation. These T cells can be rescued from apoptosis if they are provided with costimulation supplied, for example, by engaging the CD28 co-receptor with an anti-CD28 monoclonal antibody or by adding an exogenous source of interleukin-2. However, a majority of peripheral T cells from autoimmune, diabetes-prone Biobreeding (BB) rats exhibited different responses to these stimuli. T cells from these rats could not be rescued from apoptosis by costimulation. This was not due to the inability of BB-DP T cells to upregulate CD28 and the IL-2 receptor in response to TCR crosslinking. The failure of these costimulatory interactions to rescue BB-DP T cells segregated with the diabetes-susceptibility gene iddm1. Iddm1 in the rat causes peripheral T cell lymphopenia, which is associated with a dramatically shortened peripheral T cell life span. Our results indicate that a diabetogenic gene may contribute to autoimmunity by negating costimulatory signals important for the survival of long-lived peripheral T cells. © 1999 Academic Press

INTRODUCTION The Biobreeding (BB) 1 rat was derived from animals in an outbred colony of Wistar rats that displayed disease characteristics diagnosed as Type I diabetes 1 Abbreviations used: APC, antigen-presenting cells; BB, Biobreeding; DP, diabetes-prone; DR, diabetes-resistant.

0008-8749/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.



function in a regulatory capacity. The apparent lack of this population in BB-DP rats would suggest that the autoreactive effectors present in these animals could function autonomous of this protective population and thereby produce autoimmunity. The inference is that the RT6 1 T cells in the BB-DR protect the host from the RT6 2 autoreactive population present both in BB-DP and in BB-DR rats. In support of this interpretation, depletion of RT6 1 T cells from the BB-DR animal provokes markedly increased sensitivity to diseaseinducing stimuli (19) and RT6 2 thymocytes from the BB-DR can adoptively transfer diabetes (20). The prevalence of phenotypically immature T cells in the BB-DP periphery, i.e., Thy1 1 and RT6 2, suggests that autoreactivity may be associated with a stage in T cell development prior to full maturity. The way in which individual T cells perceive the antigenic universe changes as the T cell matures, but centers on the ability of the T cell to receive a signal through its antigen-specific T cell receptor, termed signal one. Subsequent signaling, usually involving an interaction with co-receptor ligands on antigen-presenting cells (APC), termed costimulation or signal two, greatly influences the T cell’s fate. In the peripheral lymphoid tissues a productive immune response normally requires both signal one and signal two. Signal one without costimulation is widely held to be an anergizing and/or death signal (21). Death is characterized morphologically as apoptosis, a process wherein the nucleus of the cell is degraded while the cell membrane remains intact (22). The impact of the developmental stage of the T cell, with respect to its fate following signaling through the T cell receptor (TCR) and costimulatory receptors, is most apparent when comparing the responses of thymocytes to those of peripheral T cells. Cortical thymocytes, as well as a subset of medullary thymocytes, are not rescued by the receipt of costimulation. Instead, two signals lead to death (23, 24). This observation has led to the interpretation that negative selection in the thymus is a two-signal process, requiring both TCR crosslinking and a second signal that can be provided by CD28. These same signals delivered to the T cell at a subsequent stage of T cell development produce activation and induce terminal differentiation into functional effector cells. It is likely that an important distinction between developmentally immature and mature T cells may be the way in which they respond to costimulatory signals delivered following TCR engagement. In the experiments presented in this paper we observed that in the absence of antigen-presenting cells the majority of purified peripheral T cells from normal rats underwent apoptosis following in vitro crosslinking with an antibody to the T cell receptor. These T cells were “rescued” from apoptotic death if an exogenous source of IL-2 was added or if a CD28-directed signal was provided by an anti-CD28 monoclonal anti-


body. However, while T cells from diabetes-resistant, nonlymphopenic animals could be rescued by costimulation, T cells from the lymphopenic BB-DP rat continued to die in the presence of costimulation provided by IL-2 or anti-CD28 antibody. Further, the failure to be rescued from apoptosis by CD28- or IL-2-mediated costimulation segregated with iddm1. These observations suggest that insensitivity to costimulation may be an important factor in the manifestation of autoreactivity. MATERIALS AND METHODS Rats Specific pathogen-free BB-DP and BB-DR rats were purchased from the University of Massachusetts Medical Center (Worcester, MA). Diabetes-resistant, specific pathogen-free Lewis (LEW) rats were purchased from Charles River (Wilmington, MA). Lymphopenic BB-DP 3 Fischer (F344) F2 animals (8) and nonlymphopenic BB-DR rats were bred at the R. H. Williams Laboratory, University of Washington (Seattle, WA) and shipped to Denver at 21 days of age. Lymphocyte donors were all used at 28 – 40 days of age. Monoclonal Antibodies The following mouse monoclonal antibodies to antigens on rat lymphoid cells were purchased from Pharmingen (San Diego, CA): R73 (directed against a constant determinant of rat TCR ab) (25), G4.18 (antiCD3) (26), OX39 (anti-IL2 receptor a chain) (27), and JJ319 (anti-CD28) (28). Apoptosis Rescue Assay Cervical and mesenteric lymph nodes were harvested from young male rats. Lymph nodes were homogenized to single cell suspensions using modified tissue grinders as described previously (29). Macrophages were depleted by panning on uncoated Falcon plastic petri dishes for 1 h at 37°C. B cells were subsequently removed by panning on anti-rat immunoglobulin (Sigma, St. Louis, MO)-coated plates (100 mg/ml) in a 5% CO 2 environment for 30 min on a series of three successive plates at 25°C. Efficiency of APC removal was assessed by FACS analysis of cells stained with anti-TCR and anti-CD3 monoclonal antibodies and was greater than 90% purified T cells. This purified population was then further enriched for T cells as described below. Six-well plates were incubated with R73 (anti-rat TCR) ascites at a 1:200 dilution (optimal concentration determined by titration) in PBS for 12 h at 37°C. Wells were washed twice with PBS and then were incubated for 1 h at 37°C with Iscove’s medium supplemented with 2% FCS to quench uncoated plastic. Enriched T



lymphocytes (5 3 10 6) (depleted of APC and B cells as described above) resuspended in 1 ml Iscove’s medium supplemented with 5% FCS were added to each well and subsequently incubated at 4°C for 1 h. The plates were gently agitated every 15 min during the 4°C incubation to disperse the cells into a uniform monolayer. The supernatant containing nonadherent cells was removed, and the monolayer was washed twice with Iscove’s to remove loosely adherent cells. Four milliliters of Iscove’s medium supplemented with 10% FCS, Pen/Strep, glutamine, and 2-ME was added to each well. This medium was additionally supplemented with nothing, 10 units/ml recombinant human IL-2 (Sigma), or 1 mg/ml anti-rat CD28 monoclonal antibody. Cells were then incubated at 37°C in 5% CO 2 until harvested for flow cytometric analysis at 48 h. Flow Cytometry Cells were activated with immobilized R73 antibody in the presence or absence of costimulatory signals as described for the apoptosis rescue assay. Approximately 1 3 10 6 cells recovered from the above apoptosis rescue assay were double-stained with either directly fluorochrome-conjugated or biotinylated monoclonal antibodies for 30 min at 4°C, washed three times, and incubated with streptavidin PE (1:400) for 20 min at 4°C. Cells were washed three times and resuspended in FACS buffer containing 6 mg/ml 7-aminoactinomycin D (7-AAD) to detect and quantitate apoptotic populations. Cells were incubated for 20 min at 4°C in the dark and analyzed by FACS on a Coulter EPICS Elite Flow Cytometer (Coulter Corp., Hialeah, FL) without removing the 7-AAD from the cell suspension. The death index was calculated by taking the ratio of dead cells (7-AAD 1) to live cells (7-AAD 2) bearing a particular phenotypic marker (i.e., percentage dead CD3 1 cells/percentage live CD3 1 cells), with dead cells exhibiting uptake of 7-AAD and live cells excluding it. RESULTS Peripheral T Cells from Normal but Not Autoimmune Donors Are Rescued from Apoptosis by Costimulation We previously published that T cells from diabetesprone BB rats can proliferate and clonally expand following polyclonal activation with the T cell mitogen concanavalin A or following stimulation with R73, a TCR-specific monoclonal antibody (30). Those studies were performed with T cell populations containing antigen-presenting cells. In this study we found that T cells in APC-depleted cultures responded differently. After 48 h of culture in the presence of plate-bound anti-TCR antibody there was no clonal expansion of purified T cells from normal LEW or diabetes-prone BB rat donors (Fig. 1A, column 1), as defined by an in-

FIG. 1. (A) Clonal expansion of purified T cells from LEW but not BB-DP cultures following the addition of costimuli. Peripheral lymph node lymphocytes (3 donors were pooled in each experiment for LEW and 10 for BB-DP) were depleted of B cells and non-immunoglobulin-bearing antigenpresenting cells as described under Materials and Methods. Purified T cells were stimulated by plate-bound R73, an antibody to the rat T cell receptor. The data in A are presented as the total number of live cells remaining 48 h after culture initiation where the starting cell number was 5 3 10 6. TCR in column 1 refers to stimulation with an antibody to the TCR alone. TCR/COS in column 2 refers to stimulation with TCR antibody in the presence of costimulation (COS). Costimulation conditions included soluble antibody to CD28 (1 mg/ml) or an exogenous source of recombinant IL-2 (10 units/ml). No clonal expansion was observed in cultures of unstimulated lymphocytes. For more detail please see Materials and Methods. Data are representative of three experiments. (B) Peripheral T cells from normal but not BB-DP donors are rescued from apoptosis by costimulation. The data from experiments described in A are reconfigured to assess the relative number of dead versus live (death index) cells following stimulation for 48 h in culture as described in A and Materials and Methods. Dead versus live cells were distinguished by the ability of live cells to exclude the dye 7-AAD. Three-color FACS analysis was used for phenotypic analysis and apoptosis quantitation. The death index represents the ratio of dead (7-AAD 1) CD3 1 to live (7-AAD 2) CD3 1 T cells in culture. Statistical analysis of death index treatment means from lymphopenic versus nonlymphopenic T cells demonstrates a significant (P , 0.01) difference between these groups when treated with COS, but not with TCR only.

crease in viable cell yields. Presented another way, as a ratio of dead/live cells in the culture, the ratio reproducibly indicated greater than 40% apoptosis within


cultures of purified T cells subjected to TCR stimulation in the absence of costimulation (Fig. 1B, column 1). This result is consistent with previous data reporting that murine peripheral T cells fail to be activated and instead apoptose when exposed to only one signal, in this case stimulation of the TCR (31, 32). In this regard T cells from normal and diabetes-prone animals appeared comparable. Based on previous data, it should be possible to rescue these T cells from apoptosis by providing second signals normally derived from the T cell interaction with an APC. To examine this, purified T cells were exposed to the R73 anti-TCR monoclonal antibody concomitant with costimulation in the form of an exogenous source of IL-2 or ligation of the CD28 co-receptor (designated in the figures as COS). When the COS second signal was provided in this way, T cells from normal LEW rats were rescued from apoptotic death. This could be demonstrated in at least two ways. First, the number of viable cells increased on average fourfold compared to cultures stimulated via the TCR alone (Fig. 1A), and second, the death index decreased (Fig. 1B). Both LEW and BB-DP APC-depleted T cell cultures subjected to TCR stimulation alone responded comparably, i.e., neither T cell population appeared to respond in a positive way to TCR stimulation alone as defined by clonal expansion. The addition of IL-2 or anti-CD28 monoclonal antibody produced very different results in BB-DP versus non-BB T cell donors. T cells from BB-DP donors were not rescued and continued to die (Fig. 1B, column 2) and there was no significant clonal expansion defined by cell yields (Fig. 1A, column 2). The experiments presented in Figs. 1A and 1B indicate that a majority of T cells from normal but not BB-DP donors can be rescued from death and induced to clonally expand by the provision of costimulation. The Role of Iddm1 in the Rescue Phenotype The BB-DP and BB-DR sublines appear to differ in diabetogenicity only in iddm1. Accordingly, the BB-DR does not display peripheral T cell lymphopenia. LEW rats differ from the BB-DP by at least two diabetogenic genes, iddm1 and iddm2 (8). Iddm2 in the rat is the major histocompatibility complex (MHC). LEW bears the RT-1 l MHC while BB-DP and BB-DR express RT-1 u. If the inability to be rescued by costimulation segregates with iddm1, then LEW and BB-DR should share the same rescue phenotype, as both strains are nonlymphopenic. Therefore the rescue phenotype of the BB-DR was examined. The results indicated that the BB-DR T cells were rescued, i.e., the BB-DR and LEW shared the same rescue phenotype and were distinct from the BB-DP (Fig. 2). If T cells from nonlymphopenic animals share the same rescue phenotype, a result consistent with the hypothesis that the failure to rescue T cells from apop-


FIG. 2. T cells from the nonlymphopenic BB-DR are rescued from apoptosis by costimulation. The experiment in Fig. 1 was repeated to compare T cells from nonlymphopenic BB (BB-DR) and nonlymphopenic/non-BB (LEW) strains (n of 3, performed twice).

tosis by costimulation is directly or indirectly controlled by iddm1, then one would predict that lymphopenic, non-BB rats should share the same rescue phenotype as the lymphopenic BB-DP. To examine this experimentally, BB-DP rats were bred with strain F344 rats. Strain F344 is compatible to strain LEW at both the Class I and Class II MHC loci, but differs from LEW in that it bears an additional diabetes resistance gene. When strain F344 3 BB-DP F2 animals are derived by mating F1’s, only 1 in 64 develops diabetes even though 1 in 4 is lymphopenic (8). The incidence of diabetes in LEW 3 BB-DP F2 animals is 1 in 16, with 1 in 4 lymphopenic. Therefore, the choice of the F344 3 BB-DP F2 combination was designed to reduce the chance that the results could be influenced by using animals with impending diabetes, an unlikely event with the F344 background. T cells from nondiabetic but lymphopenic F344 3 BB-DP F2 rats, homozygous for the diabetes-susceptibility gene iddm1 and also for the diabetes-resistant F344 MHC, displayed the same characteristics in response to TCR stimulation as did lymphopenic and diabetes-prone BB-DP rats. The results clearly demonstrated that T cells from these lymphopenic F2 animals, homozygous for a nondiabetogenic MHC, were nonrescueable by costimulation (Fig. 3). Lymphopenic F2 animals heterozygous for the BB and F344 MHC responded comparably to the lymphopenic F2 animals homozygous for the F344 MHC (data not shown). Strain F344 and nonlymphopenic F2 animals responded comparably to normal LEW animals and were rescued from TCR-induced apoptosis by costimulation. PCR genotyping confirmed that there was no influence on the rescue phenotype in animals that were heterozygous for iddm1 (data not shown). These data indicate that the failure to be rescued from apoptosis by



FIG. 3. The nonrescueable phenotype segregates with iddm1. T cells from lymphopenic F344 3 BB-DP F2 animals (n of 3) were activated by plate-bound anti-TCR for 48 h, and assessed for the capacity to be rescued by IL-2- and CD28-mediated costimulation. Rescue is defined by a decrease in the ratio of dead to live cells (death index) as determined by flow cytometric analysis of 7-AAD uptake.

IL-2- and CD28-mediated costimulation segregated with iddm1 homozygosity. BB-DP T Cells Upregulate CD28 and IL-2 Receptor in Response to Tcr Crosslinking When the TCR is engaged, costimulatory receptors on the cell surface are upregulated. These include the IL-2R and CD28. The inability of BB-DP T cells to respond positively to costimulation could therefore reflect an inability of BB-DP T cells to upregulate either CD28 or the IL-2R. This explanation appears unlikely as both CD28 and the IL-2R expression increase equivalently on both BB-DP and LEW strain T cells, demonstrated by similar increases in the mean fluorescence intensity of stimulated cells stained with antiIL-2R and anti-CD28 antibodies (Table 1 and Fig. 4).

TABLE 1 Donor


% Cells IL-2R 1

% Cells CD28 1


Unstimulated TCR Unstimulated TCR

10 91.7, 89.9 a 14 85, 88.4 a

79.9 92.6, 94.4 a 72 87.4, 74.9 a

Note. BB-DP T cells upregulate IL-2R and CD28 following in vitro TCR stimulation. Purified BB-DP and LEW T cells were stimulated for 48 h with plate-bound anti-TCR antibody as described. Cells were stained with fluorochrome-conjugated antibodies to IL-2R and CD28. Increased levels of IL-2R and CD28 proteins were demonstrated by FACS analysis in both LEW and BB-DP stimulated cultures. a Data are representative of two experiments.

FIG. 4. BB-DP and LEW T cells exhibit increased CD28 expression in response to TCR stimulation. The histograms shown illustrate the levels of CD28 expressed on LEW and BB-DP T cells that are unstimulated (Unstim) or treated with immobilized anti-TCR for 48 h (TCR), as described under Materials and Methods. The numbers on the histograms denote the mean fluorescence intensity of T cells stained with an anti-CD28 antibody, as calculated by Coulter EliteExpo Analysis software.

The Nonrescueeable Phenotype is not a Generic Characteristic of All Thy1 1 Recent Thymic Emigrants In contrast to mouse T cell development, mature T cells in the rat do not express the Thy1 antigen; instead, it is expressed on thymocytes and recent thymic emigrants. There is a marked increase in the frequency of Thy1 1 T cells in the peripheral T cell pool in the BB-DP rat (17). The exact percentage varies with age and strain, but in general terms the range of Thy1 1 T cells in normal rats is 10 –20% while in the BB-DP it is 50 –70%. It is possible that BB-DP costimulatory resistance is due to the prevalence of Thy1 1 T cells in the peripheral pool and that the nonrescueable phenotype is a generic characteristic of the Thy1 1 subset. This phenotype of Thy1 1 cells could be masked in LEW animals as a consequence of the minority of this population in the peripheral pool. To assess the contribution of Thy1 1 T cells to the rescue phenotype, Thy1 1 T cells from LEW and BB-DP rats were examined following TCR stimulation with or without costimulation. Apoptosis among the Thy1 1 subset was quantitated by three-color flow cytometry as described under Materials and Methods. The results demonstrated that most LEW Thy1 1 T cells were rescued from apoptosis by costimulation, whereas a majority of BB-DP Thy1 1 T cells were not (Fig. 5).


FIG. 5. Thy1 1 T cells from diabetes-resistant LEW but not diabetes-prone BB-DP rats are rescued by the provision of costimulation. Purified T cells were activated by plate-bound anti-TCR antibodies with and without COS, as described in Fig. 1. After 48 h in culture Thy1 1 T cells were assessed for apoptosis by flow cytometric analysis of 7-AAD uptake as described under Materials and Methods. The death index represents the ratio of dead (7-AAD 1) Thy1 1 to live (7-AAD 2) Thy1 1 T cells in culture.

DISCUSSION The experiments presented in this paper characterized T cell subsets from diabetes-prone and diabetesresistant animals to determine if distinctive phenotypes could be discerned in response to activating signals. The overall working hypothesis was that T cells from diabetes-prone BB-DP donors might escape thymic or peripheral tolerance mechanisms because they differed from T cells from normal rat donors in the way in which external stimuli were perceived or transduced to the T cell. A striking difference between T cells from BB-DP and normal rat donors was the failure of costimuli (neither an exogenous source of IL-2 nor ligation of CD28) to rescue BB-DP peripheral T cells from apoptosis (Figs. 1A and 1B). By breeding the diabetes-susceptibility gene iddm1 onto a nondiabetogenic background we observed that the failure of IL-2- and CD28-mediated costimuli to rescue T cells from apoptosis segregated with the diabetogenic gene iddm1. Iddm1 is necessary for spontaneous diabetic disease in the BB-DP rat (8); nonlymphopenic BB-DR rats, differing in diabetogenicity only at iddm1, displayed the same phenotype as LEW rats, i.e., they were rescued from apoptosis by costimulation (Fig. 3). Therefore, spontaneous autoimmunity in the BB-DP rat may be associated with the failure of BB-DP T cells to respond to costimulation provided by IL-2R or CD28 engagement. The inability of costimulation to rescue T cells from apoptosis could not be attributed to a failure of BB-DP T cells to upregulate CD28 or the IL-2R following TCR engagement. Both receptors were expressed in low lev-


els on freshly isolated lymphocytes and appeared to be equivalently upregulated regardless of the status of the T cell donor, i.e., BB-DP T cell surface phenotype by these criteria can be comparable to T cells from diabetes-resistant donors (Table 1 and Fig. 4). Therefore, the events leading to the inability of the T cells from BB-DP animals to respond to costimulation would likely occur downstream of the engagement of CD28 or IL-2R. We have reported that BB-DP T cells respond poorly to alloantigen (10) and that this response was not rescued by the provision of an exogenous source of IL-2. T cells from the CD28-knockout mouse (33) also are reported to respond poorly to alloantigen (34). Therefore, a likely explanation for the poor proliferative response to cell-bound alloantigen rests at least in part with the inability of BB-DP T cells to effectively receive CD28-directed costimulatory signals. The capacity of BB-DP T cells to proliferate vigorously in response to Con A- or anti-TCR-directed stimulation may indicate that the nature of these signals is so powerful that costimulation is not necessary. However, we favor another interpretation, namely that Con A- and TCR-directed responses could be provided by costimuli other than CD28. In support, Ramanathan and colleagues (35) reported that BB-DP lymphocytes containing both T lymphocytes and APCs expressed what appeared to be normal levels of bcl-x. Our previous data also indicate that BB-DP T cells can clonally expand in the presence of APCs, i.e., BB-DP T cells can be rescued from apoptosis if APCs are not depleted (unpublished observation). Therefore it is likely that costimulatory rescue signals can be generated by the interaction of BB-DP T cells with APCs; however, it remains to be determined to which costimulatory interactions BB-DP T cells can respond. There is good evidence that T cells from the BB-DP rat can be subdivided into at least two subsets based on lifespan. A majority of BB-DP T cells are extremely short-lived, surviving in the periphery less than 1 week (17). They express high levels of Thy1 and low levels of RT6. Adult thymectomy eliminates this population (5, 36, 37) and in so doing dramatically decreases the incidence of diabetes (5). These observations suggest that the Thy1 1 recent thymic emigrant pool in the BB-DP rat may contain autoreactive T cells. These Thy1 1 T cells also display a resistance to CD28-mediated costimulation that segregates with the iddm1 lymphopenia gene. Therefore, an iddm1-encoded failure to respond to CD28-mediated costimulation could genetically predispose the Thy1 1 T cell to become autoreactive. Our observation that Thy1 1 T cells from nondiabetic LEW rats are not resistant to CD28 signaling could be taken as evidence in support of this interpretation. Alternatively, the lymphopenia gene may simply retard T cell development beyond the recent thymic em-



igrant stage, thus unmasking an intrinsically autoreactive population that is present in nonlymphopenic, as well as lymphopenic, rats. Support for this concept comes from several sources. First, autoreactivity can be generated in animals with nonautoreactive backgrounds (38), thus demonstrating that autoreactive T cells exist in animals with no obvious predisposition to autoimmunity. Second, the BB-DR rat is not lymphopenic (8) but various inductive stimuli can induce disease (5). Finally, BB-DR thymocytes from unmanipulated animals are capable of transferring autoimmunity into athymic recipients (20). When the CD28-knockout was crossed with the diabetes-prone NOD mouse and followed for the development of diabetes it was determined that disease was exacerbated (34). This was a somewhat surprising result because treatment of the NOD mouse with antibody to CD28 protected against disease (39). One way to explain these apparently contradictory findings is to propose that the CD28-mediated signal is defective in the NOD and that the antibody treatment actually enhanced the CD28 signal transduced to T cells rather than inhibiting signaling through this molecule. Thus the complete negation of this signal with the knockout would prevent the positive effects of CD28 interactions, such as the ability to generate protective TH2 cells, as suggested by the authors. In essence, the BB-DP rat could be considered to function as a natural CD28-knockout. However, the CD28-knockout mouse is not lymphopenic, prompting the question of the relationship, if any, between the failure to respond to costimulation and lymphopenia. Recent observations by Brocker (40) provided evidence that maintenance of peripheral T cell survival required an interaction with dendritic cells. Data from Kirberg et al. (41) led to the conclusion that the peripheral maintenance process was a MHC-restricted event. In general terms, the two studies pointed to a requisite peripheral selection step to maintain T cell survival. While it was not formally established that the peripheral selection required costimulation, the peripheral lymphopenia in the BB-DP rat may be due to a failure of the BB-DP T cells to receive costimulation-dependent peripheral maintenance signals. Without them, it is clear that the murine T cells can have a short lifespan in the periphery. Although lymphopenia in the BB-DP rat can be attributed in part to a rapid death of recent thymic emigrants (5, 36, 37), the real cause of the lymphopenia may be a passive rather than an active process. Iddm1 may not be directly targeting T cells for death but rather inhibiting signals required to keep the cells alive once they reach the peripheral lymphoid organs. The death of regulatory T cells or the inhibition of their development may be a critical means whereby autoimmunity is promoted. However, a defect in CD28 signaling may also have a more direct effect on autoimmu-

nity. Recent data suggest that CD28-deficient mice also display impaired negative selection (42). ACKNOWLEDGMENTS This work was supported by Grant DK48805 from the NIH and by grants from the Juvenile Diabetes Foundation, International, and the American Diabetes Association. We thank David Stenger for his tremendous help in preparing the manuscript.

REFERENCES 1. 2. 3. 4. 5. 6.

7. 8. 9. 10. 11. 12. 13. 14. 15.

16. 17. 18. 19.

20. 21. 22. 23. 24.

Nakhooda, A. F., Like, A. A., Chappel, C. I., Murray, F. T., and Marliss, E. B., Diabetes 26, 100, 1977. Guberski, D. L., Butler, L., Kastern, W., and Like, A. A., Diabetes 38, 887, 1991. Thomas, V. A., Woda, B. A., Handler, E. S., Greiner, D. L., Mordes, J. P., and Rossini, A. A., Diabetes 40, 255, 1991. Like, A. A., Guberski, D. L., and Butler, L., J. Immunol. 136, 3254, 1986. Like, A. A., Am. J. Pathol. 136, 565, 1990. Guberski, D. L., Thomas, V. A., Shek, W. R., Like, A. A., Handler, E. S., Rossini, A. A., Wallace, J. E., and Welsh, R. M., Science 254, 1010, 1991. Markholst, H., Eastman, S., Wilson, D., Andreasen, B. E., and Lernmark, Å., J. Exp. Med. 174, 297, 1991. Jacob, H. J., Pettersson, A., Wilson, D., Mao, Y., Lernmark, Å., and Lander, E. S., Nat. Genet. 2, 56, 1992. Jackson, R., Rassi, N., Crump, T., Haynes, B., and Eisenbarth, G. S., Diabetes 30, 887, 1981. Bellgrau, D., Naji, A., Silvers, W. K., Markmann, J. F., and Barker, C. F., Diabetologia 23, 359, 1982. Elder, M. E., and Maclaren, N. K., J. Immunol. 130, 1723, 1983. Greiner, D. L., Handler, E. S., Nakano, K., Mordes, J. P., and Rossini, A. A., J. Immunol. 136, 148, 1986. Groen, H., van der Berk, J. M., Nieuwenhuis, P., and Kampinga, J., Thymus 14, 145, 1989. Lubaroff, D. M., Greiner, D. L., and Reynolds, C. W., Transplant. Proc. 11, 1092, 1979. Angelillo, M., Greiner, D. L., Mordes, J. P., Handler, E. S., Nakamura, N., McKeever, U., and Rossini, A. A., J. Immunol. 141, 4146, 1988. Hosseinzadeh, H., and Goldschneider, I., J. Immunol. 150, 1670, 1993. Zadeh, H., Greiner, D. L., Wu, D. Y., Tausche, F., and Goldschneider, I., Autoimmunity 24, 35, 1996. Rossini, A. A., Mordes, J. P., Pelletier, A. M., and Like, A. A., Science 219, 975, 1983. Greiner, D. L., Mordes, J. P., Handler, E. S., Angelillo, M., Nakamura, N., and Rossini, A. A., J. Exp. Med. 166, 461, 1987. Whalen, B. J., Rossini, A. A., Mordes, J. P., and Greiner, D. L., Diabetes 44, 963, 1995. Schwarz, R. H., Cell 71, 1065, 1992. Cohen, J. J., Duke, R. C., Fadok, V. A., and Sellins, K. S., Annu. Rev. Immunol. 10, 267, 1992. Page, D. M., Kane, L. P., Allison, J. P., and Hedrick, S. M., J. Immunol. 151, 1868, 1993. Punt, J. A., Osborne, B. A., Takahama, Y., Sharrow, S. O., and Singer, A., J. Exp. Med. 179, 709, 1994.

COSTIMULATION AND AUTOIMMUNITY 25. 26. 27. 28. 29. 30. 31. 32. 33.


¨ nig, T., Wallny, H. J., Hartley, J. K., Lawetzky, A., and HY Tiefenthaler, G., J. Exp. Med. 169, 73, 1989. Nicolls, M. R., Aversa, G. A., Pearce, N. W., Spinelli, A., Berger, M. F., Gurley, K. E., and Hall, B. M., Transplantation 55, 459, 1993. Tellides, G., Dallman, M. J., Kupiek-Weglinski, J. W., Diamanstein, T., and Morris, P. J., Transplant. Proc. 19, 4231, 1987. ¨ nig, T., J. ImTacke, M., Clark, G. J., Dallman, M. J., and HY munol. 154, 5121, 1995. Bellgrau, D., J. Exp. Med. 157, 1505, 1983. Bellgrau, D., Redd, J. M., and Sellins, K. S., Diabetes 43, 47, 1994. Parijis, L. V., Ibraghimov, A., and Abbas, A. A., Immunity 4, 321, 1996. Kishimoto, H., and Sprent, J., J. Exp. Med. 185, 263, 1997. Shahinian, A., Pfeffer, K., Lee, K. P., Kundig, T., Kishihara, K., Wakeham, A., Kawai, K., Ohashi, P. S., Thompson, C. B., and Mak, T. W., Science 261, 609, 1993. Lenschow, D. J., Herold, K. C., Rhee, L., Patel, B., Koons, A., Qin, H-Y., Fuchs, E., Singh, B., Thompson, C. B., and Bluestone, J. A., Immunity 5, 285, 1996.

35. 36.


38. 39.

40. 41. 42.


Ramanathan, S., Norwich, K., and Poussier, P., J. Immunol. 160, 5757, 1998. Sarkar, P., Crisa, L., McKeever, U., Bortell, R., Handler, E., Mordes, J. P., Waite, D., Schoenbaum, A., Haag, F., Koch-Nolte, F., Thiele, H.-G., Greiner, D. L., and Rossini, A. A., Autoimmunity 18, 15, 1994. Gold, D. P., Shaikewitz, S. T., Mueller, D., Redd, J. R., Sellins, K. S., Pettersson, A., Lernmark, Å., and Bellgrau, D., Autoimmunity 22, 149, 1995. Fowell, D., and Mason, D., J. Exp. Med. 177, 627, 1993. Lenschow, D. J., Ho, S. C., Sattar, H., Rhee, L., Gray, G., Nabavi, N., Herold, K. C., and Bluestone, J. A., J. Exp. Med. 181, 1145, 1995. Brocker, T., J. Exp. Med. 186, 1223, 1997. Kirberg, J., Berns, A., and Von Boehmer, H., J. Exp. Med. 186, 1269, 1997. Noel, P. J., Alegre, M-L., Reiner, S. L., and Thompson, C. B., Cell. Immunol. 187, 131, 1998.