Abortive activation precedes functional deletion of CD8⁺ T cells following encounter with self-antigens expressed by resting B cells in vivo

Abstract

InsHA mice express the haemagglutinin (HA) protein from influenza virus A/PR/8 H1N1 (PR8) as a self antigen on pancreatic islet β cells. We have utilized these mice to investigate the ability of resting B cells expressing K^d to induce self-tolerance among naive K^dHA-specific clone 4 CD8⁺ T cells. Adoptive transfer of K^dHA-peptide-pulsed resting B cells into clone 4 → InsHA recipients resulted in the activation and proliferation of clone 4 CD8⁺ T cells throughout the peripheral lymphoid tissues. Significantly, proliferation was not associated with the acquisition of T cell effector function; as evidenced by a lack of interferon-γ production and the complete absence of any autoimmune pathology even after immunization of recipient mice with PR8. These data demonstrate that resting B cells pulsed with self-epitopes can induce abortive activation of potentially self-reactive naive CD8⁺ T cells resulting in their functional deletion from the peripheral T-cell repertoire in the absence of any associated autoimmunity.

Introduction

Certain populations of dendritic cells (DC) play an important role in the induction and maintenance of self-tolerance among naive autoreactive CD8⁺ T cells in vivo.^1–6 However, the fact that DC have been shown to cross-present self-antigens and induce peripheral tolerance in situ does not necessarily justify their use per se when designing therapeutic strategies to induce tolerance among naive self-reactive CD8⁺ T cells.^7,8 Reports have emerged describing the adoptive transfer of antigen-pulsed immature DC into patients to modulate potentially self-reactive CD8⁺ T cells. However, the future success in using DC for the treatment of autoimmune diseases will ultimately depend upon the ability to overcome the technical challenges faced when trying to isolate and manipulate populations of DC that induce T-cell tolerance.⁹

An alternative approach would be to use resting B cells as antigen-presenting cells (APC) to induce self-tolerance because they can be easily isolated from peripheral blood (> 0·2 × 10⁹/l).¹⁰ On the other hand, DC constitute less than 1% of all peripheral blood mononuclear cells; therefore obtaining sufficient numbers of DC requires some degree of ex vivo culture and expansion of DC, CD34⁺ progenitors or CD14⁺ monocytes using cytokines such as granulocyte–macrophage colony-stimulating factor and interleukin-4.^11,12 In addition, any ex vivo manipulation can drastically alter DC phenotype and functions, such that they become immunizing rather than tolerizing.¹³

It has already been established that resting B cells are ineffective APC for priming naive CD8⁺ T cells in vitro because they possess little or no costimulatory activity.¹⁴ In vivo, it was found that injection of purified male murine B cells into female recipients resulted in subsequent CD8⁺ T-cell acceptance of male skin that was grafted 7 days later.¹⁵ However, it was not determined whether or not such graft acceptance was the result of CD8⁺ T-cell tolerance of the male H-Y antigen or the result of of the lack of antigen-specific CD4⁺ T-cell help, which is known to be important in the generation of H-Y-specific CD8⁺ effector cytotoxic T lymphocytes (CTL).^16–18

In another study, purified populations of peptide-pulsed resting B cells were shown to induce tolerance among naive CD8⁺ T cells that were specific for a foreign peptide.¹⁹ Induction of tolerance was indicated by the subsequent absence of peptide-specific CTL in vitro following subcutaneous challenge with peptide in complete Freund's adjuvant. Although peripheral tolerance induction following encounter with peptide-pulsed B cells was preceded by an initial expansion in the numbers of donor CD8⁺ T cells within the periphery, whether or not CD8⁺ T cells exhibited any effector CTL phenotype or function during this phase was not addressed. In marked contrast however, it has recently been demonstrated that, on their way to becoming self-tolerant, naive CD4⁺ T cells can pass through a significant effector phase characterized by the production of cytokines, the provision of help to naive CD8⁺ T cells and the induction of autoimmune pathology.²⁰

The experiments detailed in this report utilize an established murine model of peripheral self-tolerance induction of naive CD8⁺ T cells specific for a pancreatic islet β-cell-derived protein. InsHA mice express the haemagglutinin protein (HA) from influenza virus A/PR/8 (PR8), as a neo-self antigen on pancreatic islet β cells.^21,22 PR8 immunization of InsHA mice given naive clone 4 CD8⁺ T cells (clone 4 → InsHA), which recognize the dominant K^d-restricted epitope of HA (K^dHA), results in the productive activation of clone 4 T cells, as evidenced by the generation of large numbers of IFN-γ-producing effector cells throughout the peripheral lymphoid tissues, and the induction of immune-mediated diabetes.^23–26 In the absence of PR8 immunization, clone 4 CD8⁺ T cells proliferate only in the pancreatic lymph nodes (PLN) of clone 4 → InsHA recipients in response to β cell-derived K^dHA epitopes that are cross-presented by bone marrow-derived APC (J. M. Fraser and D. J. Morgan, unpublished data). Proliferation is extremely limited and does not give rise to autoimmune pathology or IFN-γ production, thus indicating that clone 4 CD8⁺ T cells undergo abortive activation resulting in the induction of tolerance.^23–26 In this report, these criteria have been examined to demonstrate the efficacy of using resting B cells pulsed with K^dHA self-peptides to induce self-tolerance among naive clone 4 CD8⁺ T cells.

Materials and methods

Mice

BALB/c mice were purchased from the breeding colony of The University of Bristol Animal Facility. Thy-1.2 BALB/c InsHA transgenic mice and Thy-1.1 BALB/c clone 4 T cell receptor (TCR) transgenic mice were generated and characterized as previously described.^21,23 All mice were bred and maintained under barrier conditions in the vivarium at the School of Medical Sciences in The University of Bristol. All procedures were carried out in strict accordance with the Home Office guidelines for the care and use of laboratory animals.

Peptide

Influenza virus A/PR/8/34 (H1N1) HA peptide, sequence: IYSTVASSL [518–526], was synthesized as previously described.²² Resting B cells were pulsed with 5 μg/ml per 10⁸ cells in serum-free RPMI-1640 medium (Invitrogen, Paisley, UK), containing 2 mm glutamine (Invitrogen), 1% (wt/v) penicillin/streptomycin (Invitrogen), 50 μmβ₂-mercaptoethanol (Sigma-Aldrich, Poole, Dorset, UK), and 10% (v/v) fetal bovine serum (Invitrogen; complete RPMI), for 1 hr at 37° in a humidified incubator with 5% (v/v) CO₂.

Virus

Influenza virus A/PR/8/34 (H1N1) was prepared as previously described.²¹ Mice were immunized intraperitoneally (i.p.) with 1200 HA units.

Thy-1.1 clone 4 CD8⁺ T cell enrichment

Nylon-wool columns were presoaked in complete RPMI for 50 min at 37°. Single cell suspensions of spleen and lymph node cells taken from clone 4 TCR transgenic mice were prepared and the splenic red blood cells were lysed by osmotic shock using ice-cold distilled H₂O. Cells were then loaded onto a nylon-wool column and incubated for 45 min at 37° in a humidified incubator with 5% CO₂. Non-adherent cells were eluted with complete medium; then negative enrichment for CD8⁺ T cells was achieved by incubating cells with anti-CD4 microbeads (Miltenyi Biotec Ltd, Bisley, Surrey, UK) and anti-major histocompatibility complex class II microbeads (Miltenyi Biotec Ltd). The unconjugated cell fraction was harvested using an LS/VS⁺ separation column (Miltenyi Biotec Ltd).

Preparation of purified resting B cells

A single cell suspension of mouse splenocytes was prepared from H-2^d BALB/c mice and then incubated with anti-CD45R-conjugated microbeads (Miltenyi Biotec Ltd). Cells were passed through an LS/VS⁺ column and the bound fraction was collected. The purity of B cells was determined after incubating cells with Cychrome-conjugated anti-CD45R monoclonal antibody (mAb; BD Biosciences, Cowley, Oxford, UK). In all cases the purity of the CD45R⁺ cells was found to be > 93% for CD45R with > 95% of all cells expressing CD19. Purified cells were also incubated with either biotinylated anti-CD80 or biotinylated anti-CD86 mAb followed by allophycocyanin-conjugated streptavidin (BD Biosciences). Subsequent flow cytometric analyses, carried out using a FACSCalibur (BD Biosciences), revealed that cells did not express CD80 or CD86.

Flow cytometric analyses of adoptively transferred CFSE-labelled, Thy-1.1 clone 4 CD8⁺ T cells

Purified naive Thy-1.1 clone 4 T cells were incubated with 2 μM 5-carboxyfluorescein, succinimidyl ester (CFSE; Molecular Probes, Eugene, OR), per ml of cells in phosphate-buffered saline (PBS) for 10 min at 37° in the dark. Cells were injected in 200 μl PBS into each recipient animal intravenously (i.v.). Analyses of recipient mice were carried out as follows: inguinal, brachial, axillary and cervical lymph nodes were excised and pooled together (PerLN); and the PLN and spleen were each collected separately. To detect adoptively transferred Thy-1.1 clone 4 CD8⁺ T cells, single cell suspensions were incubated with phycoerythrin-conjugated anti-Thy-1.1 mAb (BD Biosciences) and analysed by fluorescence-activated cell sorting.

To assess the production of IFN-γ in response to antigen, cells were incubated in complete RPMI with 1 μg/ml K^dHA peptide and 1 μl/ml Golgi-plug solution containing Brefeldin A (BD Biosciences), for 3 hr at 37°. Cells incubated without K^dHA peptide were used as negative controls (data not shown). Cells were washed and incubated with phycoerythrin-conjugated anti-Thy-1.1 mAb and then permeablized using a Cytofix/Cytoperm Plus kit (BD Biosciences), according to the manufacturer's instructions before being incubated with APC-conjugated anti-IFN-γ mAb (BD Biosciences).

Immunohistochemistry

Pancreata were excised and fixed in periodate-lysine-paraformaldehyde fixative then processed and embedded as previously described.²⁷ Non-specific binding sites were blocked by incubation with 10% (v/v) normal goat serum (Dako, Cambridgeshire, UK), in PBS, followed by incubation for 1 hr with guinea-pig mAb against mouse insulin (Dako). After washing for 10 min, sections were incubated with secondary biotinylated anti-guinea-pig (H + L) mAb (Vector Laboratories, Peterborough, UK), and then detected with streptavidin-conjugated horseradish peroxidase (Dako), together with liquid diaminobenzidine chromagen (Dako). All slides were counterstained with Mayer's haematoxylin (Sigma-Aldrich). The degree of islet infiltration was scored as either > 50% infiltration/complete destruction or peri-insulitis/< 50% infiltrated. Islets free of any cellular infiltration were scored as no insulitis.

Results

β cell-specific naive clone 4 CD8⁺ T cells proliferate following encounter in vivo with peptide-pulsed resting B cells

To determine the consequences following the encounter of naive clone 4 CD8⁺ T cells with resting B cells pulsed with K^dHA peptide in vivo, Thy-1.2 InsHA mice were injected i.v. with 5 × 10⁶ CFSE-labelled, purified naive Thy-1.1 clone 4 CD8⁺ T cells 1 day before receiving 5 × 10⁶ purified, resting B cells pulsed with K^dHA peptide, i.v. A separate group of clone 4 → InsHA recipient mice, given unpulsed resting B cells, was used as controls. Three days later the spleens, PerLN and PLN were harvested from each group of recipient InsHA mice and examined by flow-cytometry. CFSE-labelled, Thy-1.1 clone 4 CD8⁺ T cells were identified in peripheral lymphoid tissues from each group of mice. Figure 1 shows that in clone 4→InsHA mice immunized with PR8, naive clone 4 CD8⁺ T cells undergo many cell divisions throughout the peripheral lymphoid tissues as evidenced by the presence of many clone 4 CD8⁺ T cells having very low levels of CFSE in the spleen, PerLN and PLN (Fig. 1a–c).

Figure 1.

Naive clone 4 (CL4) CD8⁺ T cells proliferate in vivo following encounter with K^dHA peptide-pulsed resting B cells. Thy-1.2 InsHA mice were injected i.v. with 5 × 10⁶ CFSE-labelled purified Thy-1.1 clone 4 CD8⁺ T cells (a–i). After 24 hr mice were given 1200 HA units PR8 i.p. (a–c), 5 × 10⁶ purified unpulsed (d–f), or K^dHA peptide-pulsed B cells (g–i). Three days later peripheral lymphoid tissues were harvested, stained with phycoreythrin-conjugated anti-Thy-1.1 mAb and examined by flow-cytometry. Histograms show the level of CFSE (FL1) among Thy-1.1 (FL2) clone 4 CD8⁺ T cells obtained from spleen (a,d,g) peripheral lymph nodes (PerLN; from pooled cervical, axillary, brachial and inguinal nodes; b,e,h) and pancreatic lymph nodes (PLN; c,f,i). Autofluorescent cells were gated out in the FL3 channel. Data are representative of three separate experiments each containing four mice per group.

In clone 4 → InsHA recipient mice given unpulsed resting B cells, CFSE-labelled clone 4 CD8⁺ T cells proliferated only in the PLN for up to five rounds of division (Fig. 1f). Thus, under these conditions very few divided clone 4 CD8⁺ T cells were produced. Cell division was not observed within the spleen (Fig. 1d) or the PerLN (Fig. 1e). These data are consistent with an established profile of proliferation that occurs among CFSE-labelled naive clone 4 CD8⁺ T cells undergoing abortive activation in the PLN of unmanipulated InsHA recipient mice in response to cross-presented β cell-derived K^dHA epitopes²³ (Fraser and Morgan, unpublished data).

Analyses of lymphoid tissues from clone 4 → InsHA mice given K^dHA peptide-pulsed resting B cells revealed that, not only were many more clone 4 CD8⁺ T cells dividing in the PLN as compared to InsHA recipients given only unpulsed B cells, but clone 4 T cells were dividing in all of the other peripheral tissues examined (Fig. 1g–i). Under these conditions, unlike the productive activation that occurs throughout all of the peripheral lymphoid tissues of clone 4 → InsHA recipients in response to PR8 infection, clone 4 CD8⁺ T cells underwent no more than five rounds of division. Importantly, highly divided cells having low levels of CFSE were not present in any of the peripheral tissues examined.

β cell-specific naive clone 4 CD8⁺ T cells do not produce IFN-γ following encounter in vivo with peptide-pulsed resting B cells

A key feature of CD8⁺ effector T cells is that following interaction with cognate peptide they produce high levels of IFN-γ.²⁸ Productive activation of naive clone 4 CD8⁺ T cells in clone 4 → InsHA mice immunized with PR8 resulted in high levels of intracellular IFN-γ among those highly divided cells throughout the peripheral lymphoid tissues. However, following abortive activation of clone 4 CD8⁺ T cells in unimmunized clone 4 → InsHA mice there was a profound lack of IFN-γ production among any dividing cells.²⁵ Flow cytometric analyses were carried out to determine whether or not clone 4 CD8⁺ T cells proliferating within peripheral lymphoid tissues of clone 4 → InsHA mice, given K^dHA peptide-pulsed B cells produced IFN-γ upon restimulation with K^dHA peptide in vitro. As anticipated, analyses of clone 4 → InsHA recipients immunized with PR8 demonstrated high levels of intracellular IFN-γ production by divided clone 4 CD8⁺ T cells, in the spleen (Fig. 2a) PerLN (Fig. 2b) and PLN (Fig. 2c). In contrast, IFN-γ expression was not observed among dividing clone 4 T cells undergoing abortive activation in the PLN of control clone 4 → InsHA mice given unpulsed B cells (Fig. 2f) nor in the spleen or PerLN (Fig. 2d, e). Importantly, in clone 4 → InsHA mice given K^dHA peptide-pulsed B cells, intracellular expression of IFN-γ was not observed among dividing cells in the spleen, PerLN, or PLN (Fig. 2g–i). Taken together these data suggest that encounter with K^dHA peptide-pulsed B cells results in the abortive activation of naive clone 4 CD8⁺ T cells throughout the peripheral lymphoid tissues.

Figure 2.

Abortive activation of naive clone 4 (CL4) CD8⁺ T cells by K^dHA peptide-pulsed resting B cells does not result in IFN-γ production. Thy-1.2 InsHA mice were injected i.v. with 5 × 10⁶ CFSE-labelled purified Thy-1.1 clone 4 CD8⁺ T cells (a–i). After 24 hr mice were given 1200 HA units PR8 i.p. (a–c), 5 × 10⁶ purified unpulsed (d–f), or K^dHA peptide-pulsed B cells (g–i). Three days later peripheral lymphoid tissues were harvested, stained with phycoreythrin-conjugated anti-Thy-1.1 and allophycocyanin-conjugated anti-IFN-γ mAb and examined by flow cytometry. Histograms show the level of CFSE (FL1) versus intracellular IFN-γ (FL4) among Thy-1.1⁺ (FL2) cells obtained from spleen (a,d,g), peripheral lymph nodes (Per-LN; from pooled cervical, axillary, brachial and inguinal nodes; b,e,h) and pancreatic lymph nodes (PLN; c,f,i). Autofluorescent cells were gated out in the FL3 channel. The percentage of Thy-1.1⁺ cells which are IFN-γ⁺ is shown. Data are representative of cells from two separate experiments each containing four mice per group.

β cell-specific naive clone 4 CD8⁺ T cells do not induce autoimmunity following encounter in vivo with peptide-pulsed resting B cells

Previous studies have demonstrated that effector clone 4 CD8⁺ T cells are able to mediate autoimmune destruction of the pancreatic islet β cells of InsHA mice; as shown by the onset of diabetes. To determine if encounter with K^dHA peptide-pulsed resting B cells gives rise to effector clone 4 CTL, the levels of blood glucose were tested among the various groups of recipient InsHA mice (as described in Fig. 2). In control clone 4 → InsHA mice, immunization with PR8 resulted in hyperglycaemia within 8 days (Table 1). All mice in this group became sick, as evidenced by visible weight loss, as well as staring of the fur. Consistent with the development of hyperglycaemia among the PR8-immunized clone 4 → InsHA recipients, immunohistological examination of their pancreata revealed that many of the islets had already been destroyed, leaving few insulin-positive β cells, thus indicating that these mice had developed immune-mediated diabetes (Table 1). In marked contrast, clone 4 → InsHA mice given K^dHA peptide-pulsed resting B cells remained healthy and euglycaemic; as did control clone 4 → InsHA recipients that were given unpulsed resting B cells (Table 1). Immunohistological examination of pancreata taken from clone 4 → InsHA mice given either K^dHA peptide-pulsed, or from control clone 4 → InsHA mice given unpulsed B cells, revealed that all of the pancreata were free from any degree of insulitis (Table 1). In addition, there was uniform expression of insulin throughout all of the islets examined (data not shown). The fact that encounter with K^dHA peptide-pulsed B cells did not result in the generation of autoreactive effector clone 4 T cells therefore suggested that tolerance induction had occurred.

Table 1.

Incidence of autoimmunity among InsHA mice following adoptive transfer of naive clone 4 TCR CD8⁺ T cells¹

		Incidence (%)
Mice given:	Days post transfer	Hyperglycaemia	> 50% infiltration/ complete destruction	Peri-insulitis/ < 50% infiltration	No insulitis
clone 4 CD8+	8	100	1003	0	0
T cells + PR82
clone 4 CD8+	11	0	0	0	100
T cells + B cells4	21	0	0	0	100
clone 4 CD8+	11	0	0	0	100
T cells + KdHA- pulsed B cells4	21	0	0	0	100

¹Groups of adult InsHA mice were injected i.v. with 5 × 10⁶ clone 4 TCR CD8⁺ T cells as shown together with purified B cells pulsed with or without K^dHA peptide 24 hr later. Mice were monitored for the induction of diabetes by measuring the level of blood glucose, and for the presence of pancreatic islet infiltrates by immunohistological examination of pancreata post mortem on the days indicated.

²Mice were sick and were sacrificed for subsequent examination.

³Severe insulitis; most of the islets were destroyed.

⁴Greater than 250 islets were scored from a total of 3 mice per group for each time point.

Previous studies demonstrated that as a result of abortive activation in the pancreatic lymph nodes, following encounter with K^dHA epitopes cross-presented by bone-marrow-derived APC, naive clone 4 CD8⁺ T cells were functionally deleted from the peripheral repertoire.^23,24 To determine whether abortive activation following encounter with K^dHA epitopes expressed by resting B cells also results in functional deletion of clone 4 cells, the following experiment was performed. On day 0 groups of InsHA mice were injected i.v. with 10⁶, 10⁵, 10⁴, or 10³ purified clone 4 CD8⁺ T cells followed on days 1 and 8 by 5 × 10⁶ purified B cells which were either unpulsed or pulsed with K^dHA peptide. On day 18 all mice received 1200 HA units of PR8 and urine glucose was tested at regular intervals for a further 2 weeks. Table 2 shows that PR8 immunization of all mice that received any number of clone 4 CD8⁺ T cells together with unpulsed B cells resulted in diabetes within 2 weeks. Mice given 10⁶ clone 4 cells plus K^dHA peptide-pulsed B cells also developed diabetes within 2 weeks following immunization with PR8, yet immunization of InsHA mice given 10⁵ clone 4 cells plus K^dHA peptide-pulsed B cells resulted in diabetes in only one of four mice. Critically none of the mice given 10⁴ or 10³ clone 4 CD8⁺ T cells plus peptide-pulsed B cells developed diabetes following immunization with PR8. Together these data indicate that following encounter with K^dHA peptide-pulsed B cells in vivo, clone 4 CD8⁺ T cells undergo abortive activation and subsequent functional deletion from the peripheral repertoire.

Table 2.

Incidence of diabetes among adult InsHA mice following adoptive transfer of naive clone 4 CD8⁺ T cells¹

			Glycosuria2
Clone 4 Day: 0	B cells 1 & 8	PR8 18	21	25	28	32	35	40
1 × 106	alone	+	0/4	4/4	4/43,4
1 × 105	alone	+	0/4	4/4	4/43,4
1 × 104	alone	+	0/4	4/4	4/43,4
1 × 103	alone	+	0/4	3/4	4/4	4/43,4
1 × 106	+ KdHA	+	0/4	2/4	4/4	4/43,4
1 × 105	+ KdHA	+	0/4	1/4	1/4	1/43,4	0/3	0/35
1 × 104	+ KdHA	+	0/4	0/4	0/4	0/4	0/4	0/45
1 × 103	+ KdHA	+	0/4	0/4	0/4	0/4	0/4	0/45

¹Groups of adult InsHA mice were injected i.v. on day 0 with various numbers of purified clone 4 TCR CD8⁺ T cells as indicated, followed on days 1 and 8 by 5 × 10⁶ purified B cells that were either unpulsed or pulsed with 5 μg/ml K^dHA peptide. On day 18 all mice were immunized with 1200 HA units of PR8 and monitored for diabetes for a further 14 days.

²Urine glucose levels above 100 mmol/l.

³Blood glucose levels of > 14 mmol/l confirmed diabetes.

⁴Diabetic mice were culled.

⁵Normoglycemic.

Discussion

The aim of this study was to examine the consequences of the interaction between naive self-reactive CD8⁺ T cells and resting B cells pulsed with self-peptide in vivo. Our results show that clone 4 CD8⁺ T cells undergo widespread proliferation throughout the peripheral lymphoid tissues of InsHA recipients given K^dHA peptide-pulsed B cells. However, these dividing cells differ considerably from effector clone 4 CD8⁺ T cells that are generated by productive activation in response to influenza virus infection of clone 4 → InsHA recipients^25,26 because they lack IFN-γ production and are unable to cause insulitis. Critically, as a consequence of their encounter with K^dHA peptide-pulsed B cells, effector clone 4 T-cell precursors are functionally deleted from the peripheral T-cell repertoire. Thus, under these circumstances, proliferating clone 4 CD8⁺ T cells exhibit characteristics similar to those ascribed to clone 4 T cells that undergo abortive activation in the PLN of clone 4 → InsHA recipients following encounter with cross-presented β-cell-derived K^dHA epitopes (Fraser and Morgan, unpublished data).^23–25

Although proliferation of clone 4 CD8⁺ T cells is widespread in InsHA mice given K^dHA peptide-pulsed B cells, the number of times they divide does not exceed five, and therefore does not reach the number of divisions that is characteristic of a productive CD8⁺ T-cell response in this system.²⁵ One explanation for the lack of highly divided cells is related to the half-life of cell surface K^dHA on transferred resting B cells. It is possible that because of the recycling of K^d molecules there is insufficient K^dHA expressed to sustain more than five rounds of division. However, this explanation seems unlikely because adoptive transfer of a second cohort of K^dHA peptide-pulsed B cells into clone 4 → InsHA recipients, did not give rise to populations of highly divided clone 4 CD8⁺ T cells, characteristic of productive activation in peripheral lymphoid organs (data not shown). This result suggests that, following encounter with K^dHA peptide-pulsed B cells, there is a limited number of divisions that naive clone 4 CD8⁺ T cells are able to undergo as compared with naive clone 4 CD8⁺ T cells, which are productively activated following infection of clone 4 → InsHA with PR8; presumably these abortively activated cells also undergo apoptosis in situ.

Previous studies also demonstrated that during tolerance induction in vivo, naive CD8⁺ T cells underwent initial proliferation following encounter with foreign peptide expressed on resting B cells.¹⁹ However, whether or not such proliferation during tolerance induction resulted in the generation of effector CTL was not addressed directly. In our system, expression of the HA protein by pancreatic islet β cells allows us to examine directly the impact of widespread proliferation of naive self-reactive CD8⁺ T cells following encounter with self-epitopes presented by resting B cells. Effector clone 4 CD8⁺ T cells are fiercely diabetogenic, and InsHA mice demonstrate extreme sensitivity to effector clone 4 CD8⁺ T cells. Indeed, previous studies have shown that the productive activation of as few as 100 clone 4 CTL precursors is sufficient to cause diabetes, killing the InsHA mouse in less than 10 days.²⁴ Thus, it seems reasonable to predict that the generation of low numbers of autoreactive effector clone 4 T cells could result in detectable autoimmune pathology. However, the fact that none of the clone 4 → InsHA mice given K^dHA peptide-pulsed B cells demonstrated any insulitis suggests that clone 4 T cells proliferating in response to K^dHA peptide-pulsed B cells are incapable of infiltrating the pancreas and causing any autoimmunity.

Significantly, our results contrast with a study in which the process of peripheral tolerance induction of E^dHA-restricted TCR transgenic CD4⁺ T cells was accompanied by the acquisition of several significant effector functions following T cell encounter with HA expressed as a neo-self antigen in the lungs of C3-HA mice.²⁰ Several possibilities may explain this difference. It is conceivable that cross-reactivity of an alternative TCR expressed by these Rag-sufficient T cells may be responsible for the observed effector functions. Alternatively, perhaps the lack of any detectable effector function exhibited by proliferating clone 4 T cells is a feature unique to CD8⁺ T cells undergoing peripheral tolerance induction. Yet following transfer into C3-HA mice, an apparent effector phase was also observed among clone 4 CD8⁺ T cells before tolerance induction.²⁰ The fact that following transfer into InsHA mice, encountering K^dHA epitopes presented either by bone marrow-derived APC in the PLN²³ or by resting B cells throughout the periphery did not result in any effector function would therefore suggest that both the location of the antigen within the tissues as well as the context in which the antigen is encountered most likely influence the outcome.

It has been proposed that naive CD8⁺ T cells are generally tolerized by resting B cells because of the absence of sufficient costimulatory signalling to generate effector cells.^15,19,29 However, in our study the mechanism of tolerance induction employed by K^dHA peptide-pulsed resting B cells is unclear. The absence of clone 4 CD8⁺ effector cells could be the result of the lack of CD80/CD86 expression on resting B cells, or the lack of activation of resting B cells through CD40L³⁰ could result in the induction of T cell anergy. Other recent studies have demonstrated that the induction of CD8⁺ T cell anergy by lipopolysaccharide-treated B cells may rely upon cell-surface expression of transforming growth factor-β₁.³¹ However, although not yet tested, this may not be the case for anergy induction among naive clone 4 CD8⁺ T cells. Another possibility is that these clone 4 CD8⁺ cells suffer the same fate as clone 4 CD8⁺ T cells abortively activated in the PLN of InsHA mice; by undergoing apoptosis in situ.²⁵

The ability to induce tolerance among naive T cells has been assigned to populations of immature DC.^2,8 However, isolating and maintaining putative tolerizing populations of immature DC under experimental conditions is extremely difficult. The instability of the immature phenotype, and the hypersensitivity to mechanical manipulation that often results in DC maturation into an immunizing phenotype may ultimately prevent their therapeutic use.¹³ Using resting B cells as tolerogenic APC may therefore have important advantages over DC as they are relatively homogeneous with respect to their cell surface phenotype and can be easily isolated and purified from peripheral blood mononuclear cells.

Abbreviations

APC

antigen-presenting cell

CFSE

5-carboxyfluorescein, succinimidyl ester

clone 4

transgenic T-cell receptor recognizing K^d

CTL

cytotoxic T lymphocyte

DC

dendritic cell

HA

haemagglutinin from influenza virus A/PR/8/34 H1N1

InsHA

transgenic mice expressing the HA protein on pancreatic islet β cells

IFN-γ

interferon-γ

i.v.

intravenous

K^dHA

dominant H-2K^d-restricted HA peptide [IYSTVASSL]

mAb

monoclonal antibody

PBS

phosphate-buffered saline

PerLN

peripheral lymph nodes other than PLN

PLN

pancreatic lymph nodes

PR8

influenza virus A/PR/8/34 H1N1

TCR

T-cell receptor