Deletion is neither sufficient nor necessary for the induction of peripheral tolerance in mature CD8⁺ T cells

Abstract

Previous reports have demonstrated clonal deletion of CD8⁺ T cells during peripheral tolerance induction to tissue antigens. However, direct evidence demonstrating a causal connection between deletion and tolerance has not been reported because of model limitations in which the tissue antigens were expressed in vital organs. Thus, studies were initiated in a mouse model where expression of a membrane-bound ovalbumin fusion protein (mOVA) was driven by a prostate specific androgen regulated probasin promotor, providing restricted expression in a non-vital organ where antigen levels can be abrogated through androgen deprivation. Adoptive transfer of mOVA specific CD8⁺ T cells (OT-I) was used to assess the development of peripheral tolerance. Proliferation of OT-I cells was observed, as was partial deletion of transferred OT-I cells. Although deletion occurred, the long-term persistence of a stable level of OT-I cells was observed. Importantly, the persistent OT-I cells lost antigen responsiveness within 3 weeks of transfer. Castration resulted in loss of high-level prostate mOVA expression, with a resultant abrogation of tolerance induction, but surprisingly did not affect the deletion rate of OT-I cells. In contrast, abrogation of deletion through the adoptive transfer of OT-I cells from third generation CD95-deficient mice had no effect on tolerance induction. These data demonstrate the necessity for continued expression of tissue antigen throughout the establishment of peripheral tolerance. Furthermore, these findings demonstrate that deletion is neither sufficient nor required for CD8⁺ T-cell tolerance to tissue antigens, suggesting that regulatory events independent of deletion are necessary for peripheral tolerance induction to prostate antigens.

Introduction

Tolerization of T lymphocytes is necessary for the prevention of autoimmune disease. T-cell tolerance is driven centrally through negative selection and, for cells that escape thymic deletion, responsiveness to self is controlled peripherally by a number of mechanisms including regulatory cell-mediated inhibition, cytotoxic T lymphocyte antigen-4 (CTLA-4) stimulation, and clonal deletion.^1–6

Early studies outlined the mechanisms of peripheral tolerance in a normal homeostatic environment using adoptive transfer of naive antigen-specific T-cell receptor (TCR) transgenic T cells into mice containing defined antigens expressed under the control of tissue specific promoters.^7–9 These models presented the unique opportunity to monitor antigen-specific T-cell responses, providing data that established cross-tolerance as a primary paradigm for the development of peripheral tolerance. Studies using the rat insulin promotor (RIP) to drive expression of an ovalbumin/transferrin receptor fusion protein (mOVA) demonstrated the requirement for cross presentation and T-cell proliferation in tolerance induction.⁷ Antigen-specific T-cell proliferation ultimately resulted in deletion of the proliferating cells, which was hypothesized as the mechanism by which autoreactive CD8⁺ T cells were controlled.^2,3,10 Direct evidence demonstrating that deletion was necessary for tolerance development was not provided because of lethal inflammatory events directed toward β-islet antigens.

Subsequent studies in other models verified CD8⁺ T-cell proliferation induced by cross-presentation of autologous antigen and deletion of activated T cells further suggesting deletion to be the primary mechanism by which CD8⁺ T-cell responses are regulated in the periphery.^10–12 For example, studies using RIP-driven influenza haemagglutinin (RIP-HA) as a model antigen also demonstrated cross-tolerance after adoptive transfer of HA-reactive CD8⁺ TCR transgenic T cells into HA-expressing mice.¹² While deletion occurred in the RIP-HA model, quantitation of residual HA-reactive T cells was not the focus of the study. Instead, stimulation of antigen-specific cells and examination of cell numbers and inflammatory potential following antigenic stimulation was used to address the induction of tolerance. Transferred cells were rendered unresponsive to exogenous antigen over time at a variable rate dependent on antigen concentration, a process referred to as ‘functional deletion’.⁸

Later studies reported that RIP-driven production of the lymphocytic choriomeningitis virus glycoprotein (RIP-GP33) did not induce proliferation of adoptively transferred T cells under normal conditions, a situation referred to as T-cell ignorance.⁹ Recognition of self-antigen in the RIP-GP33 model was induced by immunization with antigen, and under these conditions T-cell activation led to either immunity or tolerance depending on the availability of costimulation.¹³ In this system tolerance was again associated with the deletion of autoreactive cells.

While the reports discussed above universally reported T-cell deletion in conjunction with tolerance induction, the relative insensitivity of the techniques used precluded definitive determination of total deletion. Further, studies utilizing the RIP-mOVA model reported that following adoptive transfer of high numbers of antigen specific CD8⁺ T cells, a small number of adoptively transferred cells remained in the periphery 6 weeks after transfer, suggesting that non-deletion mechanisms contributed to tolerance.¹⁴ In this regard several studies have demonstrated regulation of persistent antigen-specific cells as a mechanism for tolerance induction. In one model persistent endogenous CD8⁺ T cells were found to have down-regulated TCR affinity, resulting in loss of antigen recognition at a particular antigen concentration.¹⁵ In the same model activation of adoptively transferred high-affinity T cells by endogenous CD40 stimulated antigen-presenting cells (APCs) resulted in proliferation of T cells without gain of effector function¹⁶. In contrast, Ohlen et al.¹⁷ reported that while antigen-specific cells were fully capable of cytokine production and lysis of cognate antigen bearing cells following encounter with self antigen, peripheral tolerance was maintained by a defect in clonal expansion. Effector cells were generated but control of proliferation prevented accumulation of reactive CD8⁺ T cells in sufficient number to induce autoimmune damage.

The observations outlined above imply that multiple mechanisms exist for CD8⁺ T-cell tolerance induction in different tissues and/or under different conditions. However, all models detailed above demonstrated some level of deletion of high affinity CD8⁺ T cells during tolerance induction. Limitations in previous models, including an inability to modulate antigen expression and expression of the transgene in vital organs, limited the ability of investigators to interrogate the role of deletion. To examine the role of deletion in CD8⁺ T-cell tolerance induction, a model for a hormonally regulated prostate specific antigen was developed using the modified androgen regulated rat probasin promotor, ARR₂PB.¹⁸ The probasin ovalbumin expressing transgenic (POET-1) system utilized the ARR₂PB promotor to drive prostatic expression of mOVA. POET-1 mice were observed to express high levels of mOVA mRNA in the ventral and dorsolateral prostate after 5 weeks of age. The hormone sensitivity of this system provided a mechanism for modulating available prostate antigen levels, allowing examination of peripheral tolerance induction following loss of detectable antigen expression. Further, since the prostate is not a vital organ, numbers of adoptively transferred cells sufficient to allow analysis of undeleted cells could be utilized without fear of the life-threatening autoimmune reactions seen in other model systems.^5,7

Materials and methods

Mice

The prostate specific probasin promoter, ARR₂PB, was used to control expression of mOVA (obtained from Dr William Heath, The Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia).⁷ Five founder strains, designated POET-1 through POET-5 were obtained. A high mOVA-producing line, POET-1, was used for all the studies reported in the current manuscript. POET-1 mice were bred in house at the University of Iowa Animal Care unit. RIP-mOVA and OT-I mice (a gift of Dr William Heath), were maintained and used as previously described.⁷ B6.LPR, B6.GLD, C57BL/6 CD45.1, RAG-1 knockout mice were purchased from the Jackson Laboratories (Bar Harbor, ME). The B6.LPR, B6.GLD, and RAG-1 mice crossed with OT-I mice. OT-I^+/+ mice homozygous for the lpr, gld, or RAG-1 mutations were produced and maintained in the University of Iowa Animal Care unit. C57BL/6 animals were obtained from National Cancer Institute (NCI; Bethesda, MD, USA).

All Thy1.1⁺ OT-I⁺ mice were F1s from C57BL/6 Thy1.1⁺ bred to C57BL/6 OT-I⁺ mice, with the Thy1.1^+/+ parent line received from Jackson Laboratories (Bar Harbor, ME). All animals were maintained under specific pathogen free conditions. Unless stated otherwise in text animals were used between 10 and 24 weeks of age.

Immunohistochemistry

Tissues were embedded in Tissue-Tek^® OCT Compound (Sakura Finetek Europe; Zoeterwoude, the Netherlands) and flash-frozen in liquid nitrogen. Ten µm frozen sections were made for each tissue and maintained at 4° throughout the staining process. The frozen sections were fixed in acetone for 10–15 s and subsequently washed twice in phosphate-buffered saline (PBS). A 1 : 750 dilution of primary rabbit anti-chicken egg albumin antibody (Sigma, St. Louis, MO) was added for 30 min. Excess solution was then decanted followed by two washes in PBS. Next, a 1 : 250 dilution of secondary goat anti-rabbit immunoglobulin (IgG; whole molecule) antibody fluoroscein isothiocyanate (FITC) conjugate (Sigma) was added for 30 min. Excess solution was decanted and washes repeated. Slides were viewed with an Olympus BX-51 microscope using standard FITC filters. Images were captured using a SPOT RT Slider (Diagnostic Instruments; Sterling Heights, MI).

tdT-mediated biotin–dUTP nick-end labelling (TUNEL) assay

Tissues were prepared, using the ApoAlert^® DNA fragmentation kit (BD Clontech; Palo Alto, CA), as per manufacturer instructions. Briefly, tissues were extracted and immediately stored in fresh 2% formaldehyde. Paraffin embedded 10 µm tissue sections were affixed to glass slides and serially rehydrated (100% to 50% ethanol). Cells were stained using the provided TdT and fluorescently labelled nucleotides. Negative controls were tissue treated with fluorescently labelled nucleotides in the absence of TdT. Slides were viewed with an Olympus BX-51 microscope using FITC filters.

Reverse transcription–polymerase chain reaction (RT–PCR)

RNA used for RT–PCR was harvested from up to 20 mg of tissue using the RNeasy Mini Kit^® (Qiagen; Valencia, CA) and accompanying Qiashredder^® (Qiagen) and RNase-Free DNase Set (Qiagen). Yield was determined by UV-spectrophotometry. RNA was added to 50 ng of Random Hexamer/pd(N)6 (Roche, Basel, Switzerland) to a volume of 10 µl in RNase-free water. RNA and Oligo dT were heated to 65° for 5 min and immediately placed on ice. The RNA/random hexamer mix was added to 10 µl of a RT reaction mix ((2× First Strand Buffer; Invitrogen; Carlsbad, CA), 10 mm dithiothreitol (DTT), 2 mm dinucleotide triphosphate (dNTP) mix, 40 U RNasin^® (Promega Biosciences, San Luis Obispo, CA), and 200 U Superscript II^® Reverse Transcriptase (Invitrogen)) and incubated in sequence at 25° for 10 min, 50° for 50 min, 85° for 5 min, and stored at 4°.

Two µl of the RT reaction was used in a 25 µl total PCR reaction containing: 0·5 µm 5′ Ova Primer (AATGAGCATGTTGGTGCTGTTGC), 0·5 µm 3′ Ova Primer (5′-GAAACACATCTGCCAAAGAAGAGAACG-3′), 200 µm dNTP mix, 1× PCR Buffer (Roche), 2·5 U Taq Polymerase (Roche). PCR reaction conditions were in sequence as follows: 95° for 1 min, ((95° for 30 s, 55° for 30 s, 72° for 1 min) ×30), 72° for 5 min, and 4° hold.

Adoptive transfer

Spleens were isolated from OT-I^+/+ mice and prepared as a single-cell suspension using frosted microscope slides. B cells and macrophages were removed from the splenocytes by incubation with anti-CD24 (BD Pharmingen, San Diego, CA) followed by complement-mediated lysis. A representative sample was removed from each experiment to determine the percentage of OT-I cells in each sample. In all experiments OT-I made up 65–75% of total cells. For proliferation assays cells were loaded with carboxy-fluorescein diacetate succinimidyl ester (CFSE; Molecular Probes, Eugene, OR) as previously described.⁷ Briefly, following complement treatment T cells were incubated in 1 ml PBS/10⁶ cells with 5 µm CFSE at 37° for 10 min. For cell survival assays Thy1.1⁺ or RAG-1 OT-I cells were isolated. For all assays cells were washed three times with PBS, counted, and 7·5 × 10⁶ OT-I cells were injected intravenously. To assess deletion, approximately 7·5 × 10⁶ Thy1.1 × OT-I (CD45.2) and 5·0 × 10⁶ CD45.1 splenocytes were adoptively transferred into POET-1 or C57BL/6 mice. Prostate draining nodes were harvested at days 7, 35, and 70 post adoptive transfer. Lymphocytes were stained with anti-CD8, anti-Thy1.1 and anti-CD45.1. A gate was set on the CD8 lymphocyte population and the number of tetramer⁺ OT-I cells (Thy1.1⁺) was divided by the number of internal control cells (CD45.1⁺) to determine deletion, as presented in Fig. 4.

Figure 4.

Transferred cells are lost from POET-1 lymph nodes. Animals received 7·5 × 10⁶ Thy1.1⁺ OT-I cells retro-orbitally. (a) 7·5 × 10⁶ Thy1.1 × OT-I (CD45.2) and 5·0 × 10⁶ CD45.1 splenocytes were adoptively transferred into POET-1 or C57BL/6 mice. Prostate draining nodes were harvested at days 7, 35, and 70 post adoptive transfer. Lymphocytes were stained with anti-CD8, anti-Thy1.1 and anti-CD45.1. A gate was set on the CD8 lymphocyte population and the number of tetramer⁺ OT-I cells (Thy1.1⁺) was divided by the number of internal control cells (CD45.1⁺) to determine deletion. (b) POET-1 and C57BL/6 (BL/6) mice received OT-I cells. Twenty-one and 72 days after adoptive transfer lymph nodes were recovered and cells stained for CD8 and Thy1.1. Numbers of Thy1.1⁺ CD8⁺ were compared to total CD8⁺ cells at the draining node. Results are representative of six separate experiments. (c) RIP-mOVA and B6 mice received OT-I cells. Forty-two days after adoptive transfer lymph nodes were recovered and cells stained for CD8 and Thy1.1. Numbers of Thy1.1⁺ CD8⁺ were compared to total CD8⁺ cells at the draining node.

Castrations

All procedures were performed on animals anaesthetized with intraperitoneal (i.p.) injections of a xylazine/ketamine mixture. The scrotal area was swabbed with a Bentadine pad and testicles localized to the scrotum using manual manipulation. Following testicular localization a ∼1 cm incision was made in the scrotum through which testicles were removed from the peritoneum. To prevent bleeding a single suture was applied to the blood vessels leading to each testicle immediately before surgical removal of the testicle. The incision was closed using surgical staples and mice were monitored for 1 hr after full recovery from anaesthesia.

Recirculation interference

Twenty-four hr after adoptive transfer of OT-I cells, animals received 100 µg of purified MEL-14 (anti-mouse CD62L), a gift from Dr Morris Dailey, University of Iowa, via i.p. injection, as previously described.¹⁹ Mice were killed at various time points following MEL-14 treatment, and the axillary, inguinal, and prostate draining lymph nodes were examined for the presence of adoptively transferred cells.

Flow cytometry

Cell purity for OT-I preparations was determined by staining using antibodies against TCR chains Vb5, Va2, and CD8α (BD Pharmingen). For examination of adoptively transferred cell survival, cells were recovered from the spleen and various lymph nodes of each animal at time points ranging from 16 to 48 days after adoptive transfer, and stained with anti-Thy1.1-PerCP (BD Pharmingen) and anti-CD8α-phycoerythrin (PE). In proliferation assays cells were loaded with CFSE prior to adoptive transfer and stained for CD8α.

Cytotoxicity assays

Cytotoxicity assays were performed as previously described.²⁰ Briefly, mice were injected i.p. with 1 × 10⁸ p.f.u. of non-replicative type 5 adenovirus carrying the gene for mOVA (Ad5-mOva). Seven days after injection spleens were obtained and a single cell suspension prepared, followed by red blood cell lysis using ACK buffer treatment (0·15 m NH₄Cl, 1·0 mm KHCO₃, and 0·1 mm disodium ethylenediaminetetraacetic acid) and cocultured with E.G7 cells (an ovalbumin producing EL4 thymoma transfectant) at a 50 : 1 ratio. All cultures were kept at 2·5 × 10⁶ cells/ml in RPMI containing 10% fetal calf serum with 10units/ml recombinant interleukin-2. After 5 days of culture cells were used as effectors in a classic chromium release assay. Target cells consisted of E.G7 and the parental line EL4. All cells demonstrated less than 10% lysis of EL4 cells.

Results

Characterization of the POET-1 model system

Transgenic mice were produced using a construct containing an ovalbumin/transferrin receptor fusion protein driven by the ARR₂PB composite probasin promoter.^7,18 Five founders were identified that carried the transgene. Transgenic lines from each were developed and characterized for transgene expression by immunohistochemical staining for mOVA. Immunohistochemistry revealed mOVA protein specifically within the ventral and dorsolateral lobes in the prostate of four of five lines after 5 weeks of age (data not shown). Data are shown for one line designated POET-1, that was chosen for the remainder of this study based on similar mOVA expression to that observed in RIP-mOVA mice (Fig. 1a). Using 30 cycles of RT–PCR of tissue mRNA high-level expression of mOVA mRNA was observed to be restricted to the ventral and dorsolateral lobes of the prostate after 5 weeks of age (Fig. 1b). All other tissues were negative. To increase the sensitivity of mOVA mRNA detection, an additional 10 cycles of PCR amplification was added to the detection assay. Given saturating conditions 10 additional cycles will further amplify at a level of 2¹⁰. Following 40 cycles of amplification message was detected in multiple tissues including liver, thymus, kidney, and lymph node (Fig. 1c). Androgen ablation resulted in a decrease in mOVA mRNA in prostate tissue, with no loss of mRNA levels observed in other tissues from castrated mice, demonstrating the androgen independence of low-level mOVA expression (Fig. 1c).

Figure 1.

POET-1 mice express mOVA in prostate tissue. (a) Tissues were recovered from C57BL/6 (BL/6) (I and II), POET-1 (III and IV), and RIP-mOVA (V and VI) mice at 10 weeks of age. Frozen sections of the prostate (I–IV) or kidney (V, VI) were produced and stained with an anti-OVA antibody in either the presence (II, IV, VI) or absence (I, III, V) of 50 µg/ml exogenous OVA. (b) POET-1 mice were killed at 8 weeks of age. RNA was isolated from multiple tissues including brain (B), testis (T), liver (L), ventral (V) and dorsolateral (D) prostate lobes, kidney (K), and lung (Lu), and subjected to reverse transcription. The resulting cDNA was amplified with either mOVA or β-actin-specific primers as described in Materials and methods for 30 cycles of amplification. All results are representative of three separate experiments. (c) POET-1 mice received either castration or a mock surgery at 8 weeks of age. Twelve days after surgery mice were killed. RNA was isolated from tissues including normal combined dorsolateral and ventral prostate (P), castrated combined dorsolateral and ventral prostate (CP), normal kidney (K), normal thymus (T), normal lymph node (L), and lymph nodes from a castrated mouse (CL), and subjected to reverse transcription. The resulting cDNA was amplified with either mOVA or β-actin specific primers as described in Materials and methods. Tissues were amplified with mOVA specific primers for 40 cycles of amplification or amplified with β-actin-specific primers for 30 cycles to determine loading equivalency. (d) POET-1 and C57BL/6 (BL/6) mice received 7·5 × 10⁶ CFSE labelled OT-I cells via retro-orbital injection. Four days later prostate draining and peripheral lymph nodes were removed. Recovered CD8⁺ cells were examined for CFSE profile by flow cytometry. Cells recovered from C57BL/6 PDN and POET-1 peripheral lymph nodes demonstrated low levels of proliferation (I and II, respectively), while cells recovered from POET-1 PDN revealed a high proliferation rate for OT-I cells (III). Results displayed are representative of results from five experiments.

To examine the capability of POET-1 prostate mOVA to activate OT-I cells, adoptively transferred CFSE labelled OT-I cells were examined in prostate draining lymph nodes (PDN) 4 days following transfer. The sentinel lymph nodes for the prostate were defined by injection of carbon black into the ventral prostate followed 24 hr later by examination of lymph nodes for accumulation of carbon. In all animals tested, drainage was found to be restricted to the sacral and both lumbar lymph nodes. These three lymph nodes are referred to collectively throughout this study as the PDN. Proliferation, as monitored by the CFSE profiles of adoptively transferred OT-I cells, was examined 4 days after OT-I adoptive transfer and demonstrated clearly evident antigen-specific proliferation within the PDN, where 61% of the transferred cells in POET-1 mice had undergone two or more cell divisions. In contrast, proliferation in peripheral lymph nodes was minimal with only 17% of cells dividing (Fig. 1d). Only 5% of OT-I cells transferred into C57BL/6 mice had divided. Consistent with other models¹¹ OT-I activation resulted in up-regulation of CD69 without concomitant down-regulation of CD62L (data not shown). Notably, OT-I activation was dependent on bone marrow derived antigen presentation as determined by bm1(r)POET-1 and C57BL/6(r)POET-1/bm1 bone marrow chimeras, demonstrating a requirement for cross-presentation in OT-I activation following transfer to POET-1 mice (data not shown).

Antigen expression is quickly lost following castration

Previous studies demonstrated prostate involution and loss of autologous probasin-driven protein expression after androgen removal.¹⁸ Thus, utilization of the ARR₂PB promoter provided the unique opportunity to determine the effect of reduced antigen expression on the development of tolerance. To assess the impact of castration on OT-I activation and the development of tolerance, the rate of prostate involution and loss of mOVA expression was assessed over time. Using TUNEL staining, castration significantly elevated apoptosis in the dorsolateral prostate within 24 hr, resulting in a near total loss of epithelial cells within 3 days of surgery (Fig. 2a). The loss of mOVA expression also was observed using immunohistochemical staining of prostate tissue (Fig. 2b). Finally the impact of castration on mOVA mRNA levels was examined. Castration resulted in a dramatic decrease in mOVA RNA expression in prostate tissue within 48 h of castration (Fig. 1c). Thus, castration and the resulting loss of testicular androgens induced prostate involution and loss of mOVA expression, as detected by immunohistochemistry, providing a model for determining the impact of diminution of antigen expression on the development of tolerance.

Figure 2.

Castration results in rapid loss of antigen. (a) Prostate tissue was recovered from C57BL/6 mice at various time points following androgen ablation. Dorsolateral sections of prostate were examined for TUNEL⁺ cells at the indicated time points following surgery. I, Tissue from an untreated animal. II, Tissue taken 24 hr after castration. III, Tissue taken 48 hr after castration. IV, Tissue taken 168 h after castration. (b) Prostate tissue was recovered from POET-1 mice following castration, at the time points shown below. Frozen tissue sections were stained with an anti-ovalbumin antibody as described in Materials and methods. Serial sections were either stained in the presence of 50 µg/ml exogenous ovalbumin competitor or in 5% bovine serum albumin containing PBS. I, RIP-mOVA kidney; II, POET-1 Dorsolateral prostate before castration; III, POET-1 prostate 2 days after castration; IV, POET-1 prostate 7 days after castration; V, POET-1 prostate 10 days after castration; VI, POET-1 prostate before castration stained in the presence of exogenous ovalbumin. (c) POET-1 mice received either a mock surgery (I), or underwent castration (II). C57BL/6 control mice received mock surgery (III). Four days later mice received 7·5 × 10⁶ CFSE labelled OT-I cells retro-orbitally. Seventy-two hr after surgery lymph nodes were removed and CFSE profiles of CD8⁺ cells were examined to determine the level of OT-I proliferation. Results are representative of three separate experiments.

To examine the functional consequences of the loss of mOVA expression after castration, POET-1 mice were either left untreated, castrated, or underwent mock surgery. Four days after castration OT-I cells were labelled with CFSE and adoptively transferred into the POET-1 mice. Four days after transfer OT-I T-cell proliferation was observed in PDN of mock castrated POET-1 mice, with ∼89% dividing twice. In contrast, proliferation was minimal (∼27% dividing twice) in PDN of castrated animals, indicating a dramatic decrease in functional antigen presentation following castration (Fig. 2c).

OT-I cells adoptively transferred into POET-1 mice lose functional ability over time

Studies were performed to examine the long-term consequences of OT-I T-cell activation in POET-1 mice. To determine whether prostate-derived mOVA expression induced tolerance, the lytic activity of transferred OT-I T-cells was examined by priming with i.p. injections of Ad5-mOVA and measuring CTL activity at varying times after transfer. Transferred OT-I cells lost the ability to mediate lytic activity in a time dependent manner. Immunization with Ad5-mOVA 8 days after OT-I transfer yielded a reduced CTL response compared to those observed in wild-type mice. By 16 days post OT-I transfer POET-1 animals demonstrated negligible CTL responses following Ad5-mOVA challenge (Fig. 3a). Evaluation of the ability of OT-I cells from RAG knockout mice to expand during in vitro culture 21 days after transfer into POET-1 mice showed cells responding to SINFEKL specific tetramers expanded in BL/6 mice but not POET-1 mice (Fig. 3c). Accordingly, lytic activity was minimal in cells obtained from POET-1 mice.

Figure 3.

Transferred OT-I cells lose effector potential over time in an androgen dependent manner. POET-1 and C57BL/6 (BL/6) mice received 7·5 × 10⁶ OT-I cells retro-orbitally. At the time points listed following OT-I transfer, animals were challenged i.p. with 10⁸ p.f.u. of Ad5-mOVA. Splenocytes were isolated 7 days after immunization. Lysis of E.G7 targets was determined for each group after 5 days of in vitro re-stimulation with E.G7. Lysis of EL4 was 5% for all groups. Results are representative of three separate experiments. (b) Twelve days before adoptive transfer of OT-I cells, animals either underwent castration or received a sham operation. Animals were immunized with 10⁸ p.f.u. of Ad5-mOVA 21 days after OT-I adoptive transfer. Seven days later splenocytes were removed and cocultured with E.G7 for 5 days followed by a test for lysis of E.G7. Lysis of EL4 was 5% for all groups. Results are representative of two experiments. (c) POET-1 mice received OT-I cells by adoptive transfer while BL/6 mice did not. Twenty-one days after adoptive transfer POET-1 and BL/6 mice were immunized with Ad5-mOVA. Seven days later spleens were harvested (two mice/group), pooled and tested for tetramer positive cells. Cells were set up in coculture as described for (a) and 5 days later re-analysed for tetramer positive cells and also tested for lytic activity against E.G7 targets. Lysis of EL4 was 5% for all groups. (d) Ninety-six hr after adoptive transfer of OT-I cells, mice either underwent castration or received a sham operation. Twenty-one days after OT-I adoptive transfer axillary and prostate draining lymph nodes were analysed for proliferation by CFSE dilution or mice were immunized with 10⁸ p.f.u. of Ad5-mOVA. For the proliferative studies, CFSE dilution was determined in CD8⁺, H2K^b-SINFEKL-specific tetramer⁺ T cells. For the lytic assay, 7 days after immunization splenocytes were removed and cocultured with E.G7 for 5 days after which the cells were tested for lysis of E.G7. Lysis of EL4 was 5% for all groups. Results are representative of two experiments.

Castrated and mock treated animals were tested for OT-I function over time to examine the effect of antigen reduction during the observed development of peripheral tolerance. POET-1 mice received castration or a mock surgery, and then received 10⁷ OT-I cells via i.v. injection. Mice were immunized with Ad5-mOVA 21 days after OT-I adoptive transfer. Seven days after immunization splenocytes were harvested, expanded in a mOVA specific manner in vitro, and examined for CTL lytic activity. Castrated mice demonstrated strong CTL activity following immunization 21 days after adoptive transfer (Fig. 3b). In contrast, mock-treated POET-1 mice were unable to mount a response (Fig. 3b). To determine whether continuous antigen exposure is required for the development of tolerance, CFSE labelled OT-I cells were transferred into POET-1 and B6 mice 4 days prior to castration and subsequently tested for antigen induced proliferation 21 days after transfer (Fig. 3d). The data show that within the limits of the sensitivity of the assay, all transferred OT-I cells had responded to antigen, with >95% of OT-I cells transferred into intact or castrate POET-I mice experiencing two or more cell divisions whereas <10% of OT-I cells transferred into B6 mice proliferated. As predicted by the proliferative response observed, CD44 was up-regulated in both castrate and mock castrate POET-1 but not wild type B6 mice at day 21 (data not shown). Analysis of the ability of both POET-1 castrated and mock castrated mice to respond to m-OVA showed that tolerance was observed only in mock castrated POET-I mice. These data demonstrate that even though the OT-I cells were activated by antigen, tolerance required continued exposure to antigen.

Transferred OT-I cells undergo partial deletion

To examine the role of deletion in the tolerance observed in POET-1 mice, OT-I cells expressing the Thy1.1 marker were adoptively transferred and tracked by tetramer binding in POET-1 (THY1.2⁺) animals for up to 72 days. The data shown in Fig. 4(a) were derived from mice in which CD45.1 spleen cells also were injected and monitored as a poorly responsive, comparative T-cell population that is not be deleted. Assessment was performed 7, 35 and 70 days after adoptive transfer. A significant loss in tetramer responsive, Thy1.1 OT-I T-cells relative to CD45.1 cells was observed in all lymphoid organs examined from POET-1 mice when compared to the same tissues from wild type mice (Fig. 4a). In multiple experiments in which OT-I depletion relative to the indigenous CD8⁺ T-cell population deletion of adoptively transferred cells was incomplete, with a small but stable percentage of adoptively transferred cells persisting for longer than 72 days (Fig. 4b). While ∼70% of adoptively transferred cells were deleted within 21 days of transfer, cell numbers stabilized after the initial cell loss, with no further deletion observed up to 10 weeks after adoptive transfer.

Because OT-I loss from peripheral nodes could be indicative of trafficking to other tissues, histology of the prostate and various control tissues, including the bladder and kidney, were examined in all animals. POET-1 mice showed no inflammatory infiltrate in prostate tissue or other examined peripheral tissues following adoptive transfer of naïve OT-I cells (data not shown). Given the lack of inflammation in the prostate and other tissues under steady state conditions as well as a decrease in OT-I cells found within peripheral blood, the most likely cause of the lower OT-I cell number in lymph nodes and spleen was deletion of the OT-I cells.

To determine if the persistence of low numbers of adoptively transferred OT-I cells was specific to the POET-1 model system, the presence of adoptively transferred OT-I cells was examined in RIP-mOVA mice over time. As previously reported⁷ adoptive transfer of 7·5 × 10⁶ OT-I cells resulted in fatal disease in the vast majority of RIP-mOVA mice (data not shown); however, 20% of the mice survived up to 42 days after transfer allowing examination of OT-I numbers. As observed in POET-1 mice, 42 days after adoptive transfer ∼30% of transferred Thy1.1⁺ OT-I cells remained in the peripheral nodes of surviving RIP-mOVA mice (Fig. 4c).

To examine the role of antigen concentration over time on deletion rates, POET-1 and BL/6 mice were castrated and then received Thy1.1⁺ OT-I cells by adoptive transfer at various times before and after castration. Initially, deletion was monitored in POET-1 mice that were castrated 4 days after OT-I transfer. While castration was sufficient to prevent tolerization of OT-I cells in POET-1 mice, adoptively transferred OT-I cells were deleted to the same extent as that observed in intact animals (Fig. 5a). To insure that the deletion observed was not caused by the very early activation of OT-I cells, OT-I cells were transferred into POET-1 mice castrated from 12 days before to 4 days after OT-I adoptive transfer. The rate of deletion was observed to be equivalent in all groups tested; indicating that the presence of high-level prostate produced ovalbumin is not required for deletion (Fig. 5b).

Figure 5.

Castration does not affect deletion rate of OT-I cells. Animals received 7·5 × 10⁶ Thy1.1⁺ OT-I cells retro-orbitally. (a) Animals underwent castration 4 days after adoptive transfer. Lymph nodes were recovered 21 days after adoptive transfer and examined for the presence of Thy1.1⁺ CD8⁺ cells by flow cytometry. Results are representative of six separate experiments. (b) OT-I cells were transferred at various days before (+4) or after (−4 and −12) castration. Data are reported as the percentage of OT-I cells within the total CD8⁺ population. Results are representative of three separate experiments. (c) Thy1.1⁺ CD8⁺ cells recovered from the axillary lymph nodes of BL/6 and both castrated and normal POET-1 mice 21 days after adoptive transfer were stained with a H2-Kb/SIINFIKL tetramer to determine mOVA antigen specificity. Results are representative of two experiments.

Cells persisting 21 days after adoptive transfer were examined for their ability to bind the H2-K^b/SIINFEKL tetramer, to determine whether the persistent Thy1.1⁺ CD8⁺ T-cells remained capable of recognizing their cognate antigen. Persistent adoptively transferred cells were observed to be largely composed of tetramer positive CD8⁺ cells in both castrated and non-castrated POET-1 mice (Fig. 5c).

The continued deletion of adoptively transferred OT-I cells prompted examination of the effect of androgen independent production of mOVA in nonprostatic tissue, observed by RT–PCR, on activation and proliferation of CFSE labelled adoptively transferred OT-I cells, OT-I proliferation was examined over a 10-day period. While proliferation of OT-I cells in lymph nodes peripheral to the prostate was minor at the early (4 day) time point (Fig. 2), within 7 days of adoptive transfer OT-I proliferation was detected in all tested lymph nodes of both castrated and mock treated POET-1 animals (Fig. 6a). Thus, while antigen-specific OT-I proliferation in PDN observed 4 days after adoptive transfer relied on androgen-dependent antigen production, systemic proliferation was observed in both castrated and untreated POET-1 mice presumably from low level androgen-independent expression of mOVA throughout a variety of POET-1 tissues (see Fig. 1c). As OT-I cells did not lose expression of CD62L following activation it was possible that the cells that had undergone proliferation in peripheral lymph nodes could reflect recirculation of cells activated in the prostate draining nodes. To examine the potential role of recirculation to the appearance of proliferating cells in peripheral nodes, mice were treated with the anti-CD62L antibody MEL-14. Antibody mediated blockade of CD62L resulted in loss of naïve cells from all lymph nodes examined within 3 days of antibody treatment, without loss of naïve cells from the spleen, demonstrating sufficiency of the antibody treatment to prevent retrafficking of naïve cells to the lymph node (data not shown). Examination of peripheral lymph nodes after MEL-14 treatment revealed proliferation of OT-I cells in peripheral nodes, demonstrating direct antigen presentation at the site of non-prostatic lymph nodes. Through the use of bone marrow chimeras, antigen presentation within the peripheral lymph nodes was demonstrated to occur strictly through a cross-priming mechanism (data not shown).

Figure 6.

Low-level systemic androgen independent mOVA expression results in proliferation and deletion of transferred OT-I cells. (a) POET-1 mice received either castration (I) or mock surgery (II). C57BL/6 mice received mock surgery as controls (III). Twelve days later animals received 7·5 × 10⁶ CFSE labelled OT-I cells retro-orbitally. Seven days after adoptive transfer lymph nodes were removed and CFSE profiles of CD8⁺ cells were examined to determine the level of OT-I proliferation. (b) POET-1/bm-1 mice were lethally irradiated and reconstituted with bone marrow from either bm-1 (I) or C57BL/6 (II) mice. In parallel, C57BL/6 mice were lethally irradiated and reconstituted with bone marrow from either POET-1 (III) or C57BL/6 (IV) mice. Mice were allowed to rest for six weeks after irradiation, tested for presence of reconstituted populations, and then received 7·5 × 10⁶ CFSE labelled OT-I cells. Seven days later prostate draining lymph nodes were collected and examined for CFSE profile. (c) RIP-mOVA, POET-1, and C57BL/6 mice received 7·5 × 10⁶ CFSE labelled OT-I cells retro-orbitally. Seven days after adoptive transfer lymph nodes were removed and CFSE profiles of CD8⁺ cells were examined to determine the level of OT-I proliferation. Results are representative of four separate experiments.

To test whether the presence of proliferating OT-I cells in the non-draining lymph nodes was a unique characteristic of the POET-1 model system, the CFSE profile of OT-I cells transferred into the previously characterized RIP-mOVA mice was examined.^2,3,5,7,14 As previously reported⁷ proliferation was observed specifically within the renal lymph node within 3 days of adoptive transfer (data not shown). However, within 7 days of adoptive transfer non-draining lymph nodes from RIP-mOVA demonstrated the same pattern of OT-I proliferation seen in POET-1 animals, with both having approximately 65% of OT-I cells with two cell divisions (Fig. 6c).

Deletion is not required for tolerance induction

While the partial deletion of antigen reactive cells observed in the POET-1 model was not sufficient to induce tolerance to mOVA, the possibility that the loss of effector cell precursors associated with deletion could significantly affect the rate of tolerance induction. Previous reports demonstrated that OT-I cells were protected from deletion following recognition of mOVA self-antigen using early generations of mice bred onto the B6.lpr background, although the abrogation of deletion was lost after long-term breeding of OT-I B6.lpr mice.^3,21 To examine the potential role of deletion to the tolerance observed in the POET-1 model system, OT-I^+/+ mice were crossed twice with B6.lpr mice to yield homozygous lpr^+/+ OT-I cells that were utilized in adoptive transfer studies. CFSE labelled lpr^+/+ OT-I or gld^+/+ OT-I cells were adoptively transferred into POET-1 mice and quantitated by H2-K^b-SIINFEKL tetramer staining 21 days later. Adoptive transfer of lpr^+/+ OT-I cells resulted in increased accumulation of adoptively transferred OT-I cells in PDN rather than the partial deletion normally observed in POET-1 mice (Fig. 7a). In contrast, gld^+/+ OT-I cells were deleted in a manner equivalent to normal OT-I. Surprisingly, even though OT-I cells in PDN increased in number relative to normal BL/6 mice, the abrogation of deletion had no effect on tolerance induction in lpr^+/+ OT-I cells, implying that deletion is not required for tolerance induction in this model system (Fig. 7b).

Figure 7.

Abrogation of deletion does not affect the rate of OT-I tolerance induction. POET-1 and C57BL/6 mice received either 7·5 × 10⁶ OT-I cells, 7·5 × 10⁶ GLD OT-I cells or 7·5 × 10⁶ LPR OT-I cells via retro-orbital injection. (a) Twenty-one days after adoptive transfer lymph nodes and splenocytes were examined for H2-K^b-SIINFEKL tetramer staining. (b) Sixteen days after adoptive transfer mice were immunized with 10⁸ p.f.u. of Ad5-mOVA. Seven days later splenocytes were removed and cocultured with E.G7 for 5 days after which the cells were tested for lysis of E.G7. Lysis of EL4 was 5% for all groups.

Taken together these data demonstrate that tolerance induction is dependent on high-level prostate specific antigen expression during the induction period. These data further indicate that deletion is independent of high-level prostate expression of antigen and finally that deletion is neither sufficient nor necessary for mediation of peripheral tolerance in POET-1 animals.

Discussion

Many previous studies that characterized the development of peripheral tolerance demonstrated an association between tolerance development and deletion of peripheral T cells.^2,12 However, because of the relative insensitivity of cell quantification in vivo or a focus on functional analyses, the total absence of adoptively transferred cells remained uncertain. Additionally, the production of antigen in vital organs precluded the study of systemic tolerance under conditions in which antigen was modulated.^2,7 The androgen dependence of high levels of prostate specific mOVA production in the POET-1 model provides a unique system for addressing the requirements of antigen expression on tolerance induction. The rapid loss of antigen observed in the POET-1 model system following androgen ablation provides a convenient switch for examining activation, function, and tolerization of antigen-specific T cells following exposure to differing antigen concentrations.

Utilization of the androgen sensitivity of high-level mOVA expression yielded support for previous studies on the importance of antigen concentration in the induction of tolerance.^8,14 Data presented here corroborate previous reports that outlined the necessity of a concentration threshold for activation and tolerization, and a direct correlation between antigen concentration and the rate of tolerization.^8,9,14,22 Further, the data presented here corroborate a report of thymic deletion of specific T cells associated with extremely low level thymic production of a transgene, while simultaneously questioning the specificity of previously reported thymic transgene expression.²³

Our data also demonstrate that antigen production is required after the initial activation of OT-I cells. The observation that loss of antigen 4 days after OT-I adoptive transfer abrogated tolerance induction even though all detectable OT-I cells had experienced antigen activation. These data support previous studies stating peripheral tolerance induction requires continual antigen stimulation.^24,25 However, the previous studies could not assess the activation status of responding cells without amplification and they relied on in vitro selection and adoptive transfer to assess the necessity for continued antigen expression. The POET-I model provided an unique opportunity to assess reactive cells within the host through castration and the resulting loss of high level antigen expression. In addition, the data presented herein demonstrate that not only is antigen required throughout tolerance induction but also suggests that antigen concentrations sufficient for OT-I activation may induce different functional resolutions. This is illustrated by the ability of POET-1 mice to present antigen at a concentration sufficient to induce OT-I proliferation under both androgen-ablated and androgen-sufficient conditions, while only producing tolerance under androgen-sufficient conditions.

The correlation between mOVA message, prostate mOVA levels, and the kinetics of OT-I activation in PDN under androgen-deficient and -sufficient conditions suggests that the differences in relative antigen concentration observed in castrated and untreated POET-1 mice are responsible for the dichotomy in functional endpoints. Further, the blockade of CD62L clearly demonstrated systemic antigen presentation, and demonstrated that the differences in OT-I proliferation times observed between prostate draining and peripheral lymph nodes, was representative of differences in antigen production within the respective tissues.

The androgen-independent mOVA expression observed in POET-1 mice could reflect either very low level antigen production throughout all cells (a possibility inconsistent with other studies on concentration threshold^8,9,14) or a medium-high level stochastic antigen production in a small number of cells scattered throughout the body. Because POET-1 induced activation of OT-I cells is dependent on cross-presentation by APCs, under the stochastic model the delayed kinetics of OT-I activation in lymph nodes that do not drain the prostate could reflect reduced encounters between APCs and the rare antigen-producing cells in the periphery compared to the frequency with which APCs could encounter antigen-producing cells in the prostate.

While the POET-1 system provided several insights into the mechanisms associated with peripheral tolerance it is the examination of deletion in this model that provided the most surprising information. As reported in many other systems, peripheral tolerance in POET-1 mice was associated with deletion of antigen-specific CD8⁺ T cells.^2,12,26,27 However, deletion in the POET-1 system was incomplete, with a significant portion of adoptively transferred cells persisting in the peripheral lymphoid compartment up to 72 days after adoptive transfer, long after the induction of peripheral tolerance. Similarly, comparative studies showed that deletion also was incomplete in the RIP-mOVA model 42 days after adoptive transfer of OT-I cells. The continued presence of adoptively transferred cells appears to contradict a previous report of complete deletion in the RIP-mOVA model system.² However, a more recent report by this group, in which adoptively transferred cell numbers were significantly increased, demonstrated residual OT-I cells 6 weeks after adoptive transfer¹⁴ suggesting that the apparent contradictions between previous reports on the RIP-mOVA model and the results reported herein may reflect the differences in number of adoptively transferred cells.^8,12

It is currently unclear why deletion is incomplete in the POET-1 and RIP-mOVA models. While loss of responsiveness to antigenic signals could have prevented the delivery of a deletional signal, the induction of tolerance was not required for protection from deletion of OT-I cells in POET-1 mice, indicating that cells survived despite their continued ability to recognize and respond to a self-antigen. The protection from deletion of persistent OT-I cells could reflect either a true biphasic response to TCR mediated activation, with one population undergoing deletion and the other gaining protection from deletion, or alternatively could indicate differences in the activation status of some OT-I cells prior to the original adoptive transfer, with sensitivity to deletion being dependent on the activation status of the transferred cells. Also, given the apparent importance of Bim in deletion in other model systems, it is possible that up-regulation of anti-apoptotic molecules such as Bcl-2 following antigen recognition could account for the protection from deletion observed in a fraction of OT-I cells.²¹

Other studies have reported maintenance of long-lived unresponsive antigen specific cells, but tolerance in those cases was usually found to correlate with loss of surface TCR and/or CD8 expression.^22,24,28 While the presence of persistent cells may contradict some findings, the dual deletion and regulation of OT-I cells during tolerization fits well with data reported by Hawiger et al.²⁷ in which presentation of exogenous antigen by endogenous dendritic cells resulted in partial deletion and loss of antigen responsiveness in a manner very similar to that observed in the POET-1 system.

Further examination of the role of deletion in POET-1 peripheral tolerance using modulation of mOVA antigen levels revealed that deletion occurred in both castrated and normal males, despite the difference in antigen expression under the two conditions. The increased availability of self-antigen in a hormonally intact setting did not result in an increased rate of deletion, suggesting that either antigen concentration is not a limiting component of the deletion rate, which is supported by the similar proliferation rates in castrate and intact POET-I mice, or alternatively, that hormone dependent antigen was not utilized in deletion.

The finding that deletion occurred in the absence of tolerance demonstrated that deletion was not sufficient to induce systemic tolerance in POET-1 mice. However, the possibility that deletion played a necessary role in tolerance remained. In order to examine the necessity of deletion to peripheral tolerance, studies were performed using lpr^+/+ OT-I mice. As previously reported, deletion was abrogated in early generation lpr^+/+ OT-I mice following recognition of endogenous mOVA.^3,21 Abrogation of deletion resulted in accumulation of H2-K^b-SIINFEKL tetramer positive OT-I cells in secondary lymphoid organs, increasing the numbers of OT-I cells available for activation. Despite increases in OT-I cell numbers in the absence of deletion, lpr OT-I cells underwent tolerization to an extent nearly identical to that of CD95 wild-type OT-I, demonstrating that deletion is not required for tolerization in the POET-1 model. However, while deletion appears superfluous in POET-1 tolerance under these conditions, further experiments are required to determine what contribution, if any, deletion applies to peripheral tolerance induction under other circumstances, particularly those in which antigen-specific cell numbers are more reflective of physiological T-cell levels.

The observed necessity of high level antigen production throughout CD8⁺ T-cell tolerance induction correlates well with studies that outlined the antigen requirements for CD4⁺ T-cell tolerance ²⁹ as well as for the induction of CD8⁺ T-cell tolerance to both intestinally delivered antigen ³⁰ and subcutaneous peptide.³¹ The role of continual antigen expression in tolerance could indicate a necessity for continual stimulation of either a regulatory cell population or the OT-I cells themselves during tolerance induction. While the mechanisms by which T cells in the POET-1 system are regulated remains unclear, several deletion independent mechanisms of T-cell tolerance induction have been previously described.^16,17,31,32 Other studies have begun to address the role of dedicated regulatory T cells in the control of autoimmune responses.^33,34 Further studies are necessary to determine which, if any, of these mechanisms are involved in regulation of persistent OT-I cells, and perhaps most importantly what methods are capable of inducing recovery of antigen responsiveness in the persistent population.

In summary, the data presented herein demonstrate a biphasic response to a prostate antigen, characterized by partial deletion and loss of responsiveness in a persistent antigen-specific CD8⁺ T-cell population. These responses were found to be dependent of different antigen concentrations, with regulation of persistent cells requiring higher levels of antigen production than deletion. Further, this study reveals that the deletion component of the CD8⁺ T-cell response to self-antigen is neither necessary nor sufficient for tolerance induction. The results of this study suggest that regulation of persistent CD8⁺ T cells, independent of deletion, is a necessary step in the control of peripheral T-cell responsiveness.

Abbreviations

mOVA

membrane-bound transferrin-receptor/ovalbumin fusion protein

POET

prostate ovalbumin expressing transgenic mice.