Systemic lupus erythematosus is an autoimmune disease caused primarily by autoantibodies (including IgG anti-DNA) and immune complexes that cause tissue damage. After tolerization with an artificial peptide (pConsensus, pCons) based on murine anti-DNA IgG sequences containing MHC class I and class II T cell determinants, lupus-prone (NZB × NZW)F1 female (BWF1) mice develop regulatory CD4+CD25+ T cells and inhibitory CD8+ T cells, both of which suppress anti-DNA Ig production and immune glomerulonephritis. In the present work, we show that splenocytes from BWF1 mice treated with pCons had significant expansion of primarily CD8+ T cells. CD4+ T cells and B cells were each directly suppressed by CD8+ T cells from tolerized mice in a contact-independent manner. Both pCons-induced CD8+CD28+ and CD8+CD28− T cells suppressed production of anti-DNA in vitro. Silencing with small interfering RNA of Foxp3 abrogated the suppression mediated by both CD8+ T cell subsets. Additionally, CD8+ T cells from tolerized mice were weakly cytotoxic against syngeneic B cells from old anti-DNA-producing mice, but not from young mice. Importantly, pCons treatment had dual effects on CD8+ suppressor T cells from tolerized mice, increasing the intracellular expression of Foxp3 while decreasing the surface expression of PD1 molecules. Blocking PD1/PDL1 interactions in the CD8+ T cells from tolerized mice reduced their expression of Foxp3 and their ability to suppress CD4+CD25− proliferation. In contrast, blocking PD1/PDL1 in naive T cells increased Foxp3 expression. Our data suggest that tolerization with pCons activates different subsets of inhibitory/cytotoxic CD8+ T cells whose targets are both CD4+CD25− effector T cells and B cells.
The development of autoimmunity depends on polygenic predisposition and, in some cases, on environmental factors. Systemic lupus erythematosus (SLE)3 is associated with autoantibodies, including IgG anti-DNA, and immune complexes that cause organ damage (1, 2, 3). Autoreactive B cells and helper T cells, as well as impairment in regulatory and suppressive T cells, have been reported both in animal models of SLE and in lupus patients (4, 5, 6, 7). Recently, induction of regulatory/suppressive cells has been envisioned as a novel therapeutic strategy to control pathologic aggressive cells in SLE and in other autoimmune conditions (8, 9, 10, 11, 12, 13, 14, 15, 16, 17).
We have developed a treatment in female BWF1 mice using tolerogenic doses of an artificial 15-mer peptide (pConsensus, pCons) that is based on sequences containing MHC class I and MHC class II determinants in the VH region of several J558-encoded BWF1 mAb anti-DNA. Administration of the peptide in tolerogenic doses significantly delayed the onset of nephritis, decreased the serum levels of autoantibodies, reduced serum levels and CD4+ T cell secretion of IFN-γ, and prolonged survival (3, 18). The mechanism of protection depends in part on induction of anergy of CD4+ helper T cells, generation of peptide-specific CD4+CD25+ regulatory T cells (5) and activation of CD8+ inhibitory T (Ti) cells (6, 18).
Several studies have shown that CD4+ regulatory T cells and CD8+ suppressor T cells play a crucial role in the prevention of autoimmunity via several different mechanisms, including cell–cell contact, and TGFβ secretion in the case of pCons-induced CD8+ Ti cells (6, 19, 20, 21, 22, 23). Regulatory functions of CD8+CD28− T cells in tolerance and heart transplant recipients have also been described (24, 25, 26). Human regulatory CD8+CD28− T cells generated in vitro suppress the activation and proliferation of Th cells induced by allogeneic cells. This effect requires cell-to-cell contact and is Ag-specific.
We confirm here a role of CD8+ T cells in the p-Cons-induced suppression of lupus-like autoimmunity, and show that: 1) these cells are greatly expanded in tolerized mice; 2) they can directly suppress proliferation of both T and B syngeneic cells; 3) they are resistant to apoptosis; 4) a CD8+ T cell subpopulation induces cytotoxicity in anti-DNA-producing B cells; 5) both CD8+CD28+ and CD8+CD28− T cell subsets inhibit anti-DNA production; 6) inhibition can be abrogated by silencing Foxp3; and 7) the emergence of CD8+ cells is associated with down-regulation of surface expression of PD1, and inactivating PD1/PDL1 interactions abrogates suppression. Thus, the observed suppressive capacities depend in large part on expression of Foxp3 intracellularly and alteration of PD1 on the cell surfaces.
Materials and Methods
BALB/c (H-2d) and (NZB × NZW)F1 (H-2d/z) mice were bred and maintained at the University of California–Los Angeles or purchased from The Jackson Laboratory. Mice were housed in pathogen-free conditions and treated in accordance with the guidelines of the University of California–Los Angeles Animal Research Committee, an institution accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). All experiments were conducted in female mice.
The peptides used in this study and the MHC molecules they bind are described in detail elsewhere (6). The tolerizing peptide pCons (FIEWNKLRFRQGLEW) contains T cell determinants based on the J558 VH regions of several murine mAb anti-dsDNA from BWF1 mice (27). The negative control peptide (pNeg) (AIAWAKARARQGLEW) binds I-Ed (expressed by BWF1) but is nonstimulatory and nontolerogenic. Wild 12-mer or 15-mer peptides from VH of BWF1 anti-DNA Abs that stimulate CD4+ T cells from BWF1 mice include p7 (GYFMNWVKQSHGKSL), p34 (MNWVKQSHGKSL), and p58 (FYNQKFKGKATL) (6, 27). PCDR1 (TGYYMQWVKQSPEKSLEWIG) is a wild stimulatory peptide described by Eilat et al. (28) from a similar region in the VH of a murine mAb anti-DNA Ig. Other nonstimulatory control peptides are pHyHEL (VKQRPGHGLEWIGEI), derived from the CDR1/Fr2 VH region of a murine Ab against hen egg lysozyme (HEL), and p11 and p93, which derive from the same VH of the stimulatory wild Ig peptides as p7, p34, and p58 (BWF1 anti-DNA Ab A6.1) (29). Peptides were synthesized at Chiron Biochemicals, purified to single peak on HPLC, and analyzed by mass spectroscopy for expected amino acid content.
Treatment of mice
For tolerance induction, 10–12-week-old BWF1 mice received a single i.v. dose of 1 mg of one of the peptides, dissolved in saline, as reported previously (3). Controls in selected experiments were either treated with saline or control peptides.
Cell isolation and staining
Spleen cells were isolated from saline-treated, naive or tolerized, BWF1 mice 1 wk after administration of pCons (except when specified otherwise in legends) after lysis of RBC with ACK lysing buffer (Sigma-Aldrich). Cell subsets were purified by incubation with anti-CD4, anti-B, anti-CD8+, anti-NK1.1, anti-Mac-3, and anti-Gr-1 microbeads from (Miltenyi Biotec). A total of 1 × 2 × 106 freshly isolated spleen cells or CD8+ T cells were used for staining of cell-surface molecules. Abs used to analyze the cells included anti-Thy1.2, anti-CD4, anti-B220, anti-CD8, anti-CD25, and anti-CD28 (all from BD Pharmingen).
Cell sorting was performed on stained splenocytes from naive and pCons-treated mice. Splenocytes were prepared after RBC lysing, and 10 × 106 cells/ml were stained with FITC-conjugated anti-mouse CD8, APC-conjugated CD28 Abs from BD Pharmingen. Cells were sorted with FACSVantage SE flow cytometer (BD Biosciences) at the University of California–Los Angeles Flow Cytometry Core Facility.
Isolated cells were washed with FACS buffer and 1–2 million cells were used for surface staining and immunophenotyping. Before staining, cells were incubated with rat anti-mouse CD16/CD32 (FCγ III/II receptor) mAb to block nonspecific binding. Cells were then stained with Abs to anti-mouse CD3 (clone-145-2C11), CD8a (Ly-2) (53-6.7), CD4 (L3T4) (clone-RM4–5), CD45R/B220 (RA3–6B2), NK1.1 (PK136), CD49b/Pan NK (DX5), Mac-3 (M3/84), GR-1 (RB6-8C5), CD11c (HL-3), CD25 (PC61), CD28 (37.51), granzyme B (16G6), and PD1 (RMP1–30). Immunophenotyping of splenocytes from untreated and pCons-tolerized mice was performed with a FACSCalibur flow cytometer (BD Biosciences) using either CellQuest (BD Biosciences) or FCS Express software (De Novo Software). Staining with multiple combinations of Ab was performed according to standard procedures described elsewhere (5, 6, 18). Staining with annexin V and with 7-aminoactinomycin D (7AAD) was used to distinguish cells undergoing apoptosis from necrotic cells. The conjugated Abs used were purchased from BD Pharmingen and eBioscience.
For CFSE labeling, purified CD4+ T cells and B cells were adjusted to 1 × 107 cells/ml in RPMI 1640 containing 5% FBS. These cells (100 μl) were mixed and incubated with 5 μM of fluorescent dye CFSE (Molecular Probes) for 5–10 min at 37°C. The reaction was quenched with 10 ml of cold RPMI 1640 media supplemented with 5% FBS followed by washing twice with cold RPMI media containing 10% FBS. The cell division status of cells was determined by measuring CFSE fluorescence halving. The labeled cells were cultured in the presence of 10 ng/well of recombinant murine IL-2 for 3–5 days with nonirradiated and irradiated B cells together with naive CD8 and tolerized CD8+ T cells. Cells were harvested, washed, stained separately for B, CD4, and CD8+ T cells, and expression of CFSE on CD4+ T cells was analyzed by flow cytometry.
In vitro suppression assay
Spleen cells were isolated from BWF1 mice after 1 wk of pCons treatment as described before. B220+ B cells and CD4+ and CD8+ T cells were isolated via magnetic bead separation using VarioMACS apparatus (Miltenyi Biotec). CD4+ T cells, CD4+CD25− cells (as responders), irradiated and nonirradiated B cells as APCs, and CD8+ T cells (as suppressors) from pCons-treated mice were used in the experiments. Isolated cells (1 × 105) were cultured in triplicate in 96-well plates with varying amounts of CD8+ T cells, 2 × 105 irradiated and nonirradiated B cells with 20 μg/ml pCons or control peptides to activate suppression for 96 h, then pulsed with 0.5 μCi/well [3H]thymidine (PerkinElmer) during the last 18 h of culture. Cells were harvested using an automated cell harvester onto filters, and radioactivity was counted in a Beckman scintillation counter.
For intracellular staining, cells were first stained for expression of cell-surface markers and then fixed, permeabilized, and stained using the BD Cytofix/Cytoperm kit (BD Pharmingen) according to the manufacturer’s instructions.
Cytotoxicity assay was performed using the CytoTox 96 nonradioactive cytotoxity assay (Promega) as per the manufacturer’s instructions. The experiments were performed in coculture experiments of target B cells (from naive BWF1 mice of various ages) with isolated and sorted CD8+CD28− T cells from spleens of naive BWF1 mice, or from spleens of BWF1 mice tolerized 1 wk earlier with pCons, at different E:T ratios as indicated in figure legends.
Cytokine measurement in the supernatant of cultured spleen cells was done with BD OptEIA ELISA kits (BD Biosciences and BioLegend) for IFN-γ, IL-2, TGFβ, and IL-10. Intracellular mRNA encoding IFN-γ, IL-10, Foxp3, and TGFβ was analyzed by real-time RT-PCR. GAPDH was used as a housekeeping gene for normalization.
Assays for measurement of anti-DNA Ab
Assays were performed to measure anti-DNA Ab as described earlier (3, 5, 6, 18). For optimal Ab production, we cocultured B cells from old naive BWF1 females with 2+ proteinuria or higher, with CD4+ T cells from young 10–12-wk-old naive BWF1 females without proteinuria, and with CD8+ T cells from 10–12-wk-old females treated 1 wk before with saline or pCons. Ratios are 1 B cell to 10 CD4+ T cells to 10 CD8+ T cells. After 5 days, culture supernatants were collected, concentrated, and analyzed for anti-DNA IgG by ELISA.
Real-time PCR was analyzed as described earlier (6, 18). Briefly, total RNA was isolated with TRIzol (Invitrogen) as per the manufacturer’s protocol. Reverse transcription used 50 ng of total RNA. The oligonucleotide sequences used for the primers and TaqMan probes are as follows: IFN-γ forward, 5′-TGA GAC AGA AGT TCT GGG CTTCT-3′; reverse, 5′-CAAGAT GCA GTG TGT AGC GTTCA-3′: probe, 6FAM TCC TGCGGCCTAGCTCTGAGA TAMRA. IL-10 forward, 5′-CAG CCG GGA AGA CAA TAA CTG-3′; reverse, 5′-CCG CAG CTC TAG GAG CAT GT-3′; probe 6FAM ACC CAC TTC CCA GTC GGC CAG AG TAMRA. TGFβ forward, 5′-AAACGGAAGCGCATCGAA-3′; reverse, 5′-GGGACTGGCGAGCCTTAGTT-3′, probe 6FAM CCATCCGTGGCCAGATCCTGTCC TAMRA. Foxp3 forward, 5′-TGCAGGGCAGCTAGGTACTTGTA-3′; reverse, 5′-TCTCGGAGATCCCCTTTGTCT-3′; probe 6FAM TCCGAACAGCATCATCCTTCTTAGCATCC TAMRA. PD1 forward, 5′-AGGCCGCCTTCTGTAATGGT-3′; reverse, 5′-GGGCAGCTGTATGA TCTGGAA-3′; probe 6FAM AGCCAACCCGTCCAGGATGCC-3′ TAMRA. The amplification primers were at 900 nM and the probes were at 200 nM. A passive reference dye (ROX) provided an internal standard for normalization of FAM fluorescence, correcting for fluctuations due to volume changes. For relative quantitation, a standard curve was constructed for each primer and probe set, using total RNA. RNA was isolated from spleen cells of 10–13-wk-old naive or tolerized mice. Spleen cells from 2–3 mice in each group were pooled for each experimental group. For some experiments, CD4+ and CD8+ cells were isolated by positive selection using microbeads with Miltenyi AutoMACS as described above. A ribosomal RNA control primer and probe set (Applied Biosystems) were used for normalization purposes. The possibility of genomic DNA contamination was excluded by use of no-reverse-transcriptase controls in combination with ribosomal primers. GAPDH was used as an endogenous control in each experimental set. All samples were run in duplicate. Normalization was used as indicated in the figure legends.
CD8+CD28+ and CD8+CD28− T suppressive cells and CD8+ T cells isolated as described above were plated and cultured in 24-well plates for 24 h in complete medium containing 10% FCS. For transfection, a Silencer siRNA transfection kit from Ambion was used. OptiMEM reduced serum medium (Invitrogen) was used to dilute the siPORT amine. Validated siRNA of Foxp3 and GAPDH were obtained from Ambion, as well as positive and negative siRNA controls. The negative control siRNA was a scrambled sequence that bears no homology to human, mouse, or rat genomes. The transfection agent alone served as another control (siPORT amine). The agent was mixed with siRNA of Foxp3 (50–100 nM) and GAPDH (50–100 nM) or controls in serum-free medium and incubated at room temperature for 30 min. Cells were transfected with siRNA complexes by overlaying siRNA dropwise onto the cells. After 8–10 h, the medium was removed and fresh medium (1–2 ml) was added. Viability was assayed with trypan blue staining. After 48 h of culture, transfected CD8+ T cells were transferred to cultures of fresh BWF1 CD4+ T cells plus B cells plus pCons for measurement of suppression of anti-DNA Ab production. Some transfected cells were lysed with cell lysing solution (Invitrogen) and RNA was isolated for real-time PCR to validate knock down of the target gene.
Statistical analyses were performed using Prism 4 software (GraphPad Software). Parametric testing between two groups was performed by paired t test or by Mann-Whitney U test. Nonparametric testing among more than two groups was performed by one-way ANOVA (p values <0.05 were considered to be significant).
Immunophenotyping of BWF1 splenocytes after tolerization with pCons
To determine phenotypic changes after pCons treatment, flow cytometry was performed. The following are the changes in splenocytes after pCons treatment.
Expansion of CD8+ T cells after tolerization with pCons
We did flow cytometry analysis of splenocytes comparing cells from tolerized mice to cells from naive mice. pCons-treated mice had increased percentages and numbers of T cells expressing CD3+CD8+. In previous work, we had reported that the numbers of CD8+ T cells in BWF1 mice tolerized 1 wk before spleen cell harvest increased by 4-fold (6), as did the numbers of pCons-binding CD4+CD25+ regulatory T cells (5). In the work done here, FACS analysis revealed a significant associated decrease in the percentage of B cells after pCons treatment, an increase in the percentage of monocytes, and a stable percentage of cells expressing surface markers for total CD4+ T cells, NK cells, and granulocytes, but no change in total cell numbers of any of these populations (data not shown). Therefore, the only cells that expanded in numbers among those tested were the CD8+ T cells.
After administration of pCons, a population of immune cells expanded. In particular, one week after tolerization, more cells were larger in size as compared with cells from naive animals, as shown by forward scattering of cells (FSC) (Fig. 1, A and B). To know the cell subtypes present among those large cells, we stained for surface molecules (CD8+, CD4+, B220, CD11b, NK1.1, and Gr-1) and gated on large splenocytes. Most of the gated expanded cells were CD8+ T cells (Fig. 1,C). We also found significantly increased total numbers of CD8+ T cells and decreased CD4+ T cells in the enlarged gate of splenocytes from pCons-treated mice compared with naive mice (Fig. 1,D, ∗, p < 0.05). There was no significant change in total cell numbers of B cells, granulocytes, macrophages, and NK cells. pCons-treated CD8+ T cells from enlarged gate based on FSC expresses a 4–5-fold increased mRNA of Foxp3 (Fig. 1,E) and more prominent down-regulation of PD1 as measured by real-time PCR (Fig. 1 F). These findings were present between 5 and 15 days after tolerization. The data confirm large effects of tolerization on the expansion of CD8+ T cells.
Foxp3 expression is increased in CD8+ T cells after pCons treatment
We found that CD8+ T cells from pCons-treated mice had 3–4-fold higher Foxp3 mRNA compared with naive CD8+ T cells as measured by real-time PCR (Fig. 1,G). To determine whether pCons induced intracellular Foxp3 in addition to mRNA, we did intracellular staining for Foxp3. Intracellular staining revealed increased Foxp3 expression (Fig. 1,H). Mean fluorescence intensity (MFI) of Foxp3 staining in pCons-treated CD8+ T cells was significantly higher in tolerized than in CD8+ T cells from naive mice (Fig. 1,I). Foxp3 staining was different from isotype control, as shown in Fig. 1 J. We have previously reported that the numbers of pCons-binding CD4+CD25+ cells expressing Foxp3 are increased 1 wk after tolerization (5), and in the work reported herein we confirmed a statistically significant increase of CD4+Foxp3+ T cells in pCons-treated groups as compared with naive mice (data not shown). These data suggest that pCons induces Foxp3 in both CD4+ and CD8+ T cells.
Silencing with siRNA of Foxp3 abrogates the suppression of anti-DNA IgG induced by CD8+ T cells
We have previously reported that CD8+ Ti cells from tolerized mice suppress anti-DNA production by syngeneic CD4+ Th plus B cells in vitro, and this suppression can be abrogated by silencing Foxp3 (18). To determine whether this occurs in both subsets of CD8+ T cells (CD8+CD28+ and CD8+CD28− cells), we transfected sorted CD8+CD28+ and CD8+CD28− T cells from both naive and pCons tolerized mice with Foxp3 siRNA. Then, the nontransfected and transfected CD8+CD28 subsets were cultured with naive CD4 plus B cells. After 4–5 days, culture supernatant was collected and analyzed for anti-DNA IgG with ELISA. We found that both the subsets CD8+CD28+ and CD8+CD28− from tolerized mice suppressed production of anti-DNA IgG (Fig. 2,A, compare column c vs columns h and j, ∗∗, p < 0.005). Silencing with siRNA of Foxp3 in both subsets abrogated the suppression; however, abrogation was greater in CD8+CD28+ subsets of tolerized mice (Fig. 2 A, columns i and k, ∗, p < 0.05).
To determine whether CD8+CD28− T cells from pCons-treated mice had increased mRNA of Foxp3, we sorted CD8+CD28+ and CD8+CD28− subsets from pCons-treated mice splenocytes, and RNA was isolated and real-time PCR was performed with Foxp3-specific primers and probes. We found that CD8+CD28− T cells had significantly (p < 0.022) higher 2–3-fold mRNA of Foxp3 (Fig. 2 B). These data suggest that both the subsets of tolerized CD8+ cells play an important role in suppression of anti-DNA Ab, although Foxp3 mRNA expression was higher in the CD8+CD28− subset.
Polyclonal activation significantly increases mRNA expression of TGFβ, IL-10, and Foxp3 in tolerized CD8+ T cells
Because Foxp3 has been reported to play an important role in the CD4+CD25+ regulatory T cells that protect from autoimmunity (30, 31, 32, 33, 34, 35, 36), and because Foxp3 expression was consistently up-regulated >2-fold in tolerized mice CD8+ T cells, we sought to investigate the mRNA expression of Foxp3 after TCR polyclonal activation. Isolated CD8+ T cells from tolerized mice 1 wk after pCons treatment and from naive mice were stimulated and cultured for 24 h with plate-bound anti-CD3 (2 μg/ml) followed by anti-CD28 (4 μg/ml). Cells were lysed with RNA lysis solution, and RNA was isolated both from unstimulated and stimulated CD8+ Ti cells. Foxp3 expression was significantly increased in CD8+ T cells from pCons-treated animals, but it increased even more (15-fold) after polyclonal activation (data not shown and Ref. (6)).
We next examined the mRNA levels of other molecules reportedly involved in Foxp3 regulation. Without additional activation, TGFβ mRNA levels were increased 2-fold in CD8+ T cells from tolerized mice compared with naïve mice (Fig. 3,A, p < 0.043). Polyclonal activation with anti-CD3 (2 μg/ml) followed by anti-CD28 (4 μg/ml) induced TGFβ >3-fold higher in the tolerized cells (Fig. 3,B, p < 0.05 and Ref. (6)). In contrast, mRNA expression of IFN-γ and IL-10 were not increased in CD8+ T cells from tolerized mice, although both cytokines were increased more than 3-fold by polyclonal activation of the cells (Fig. 3, A and B). Therefore, it is possible that up-regulation of Foxp3 in our mouse model of SLE is linked to cytokine changes, particularly TGFβ. Recently, it has been shown that TGFβ induces Foxp3 expression and maintains suppressive function in CD8+ Ti cells and CD4+ regulatory T (Treg) cells (18, 20, 37, 38, 39, 40).
CD8+ Ti cells suppress proliferation of naive CD4+ T cells and B cells
We analyzed the target of suppression by CD8+ Ti cells and found that the targets are both CD4+ T cells and B cells. As shown in Fig. 4,A, tolerized CD8+ Ti cells mixed with CD4+ T cells plus B cells suppressed proliferation in a dose-dependent manner. Naive CD8+ T cells cultured alone proliferated better than tolerized CD8+ T cells. Tolerized CD8+ T cells in coculture suppressed the proliferation of naive CD4+ T cells plus naive B cells (Fig. 4 B) when stimulated with pCons, but not with pNeg. Thus, peptide-specific tolerized CD8+ T cells suppressed the proliferation of naive CD4+ T cells and B cells in cocultures.
Next, we labeled CD4+ T cells with CFSE, and CFSE-labeled CD4+ T cells were cultured with irradiated B cells as APCs together with CD8+ T cells from tolerized mice or naive mice. We found that tolerized CD8+ T cells suppressed the number of CD4+ T cells that were proliferating (Fig. 5,A). We also analyzed whether pCons-reactive CD8+ T cells directly suppressed B cell proliferation. B cells were labeled with CFSE and then cultured with naive CD8 or tolerized CD8+ T cells. We found that pCons-tolerized CD8+ T cells suppressed the proliferation cycles of B cells, independent of the presence of CD4+ T cells (Fig. 5, B and C). Thus, the targets of CD8+ Ti cells are both CD4+ and B cells. Transwell experiments indicated that the CD8+ Ti cells might not need contact for suppression of proliferation of either cell type (data not shown). To determine the role of TGFβ in the suppression assay, we did the blocking experiments with anti-TGFβ Ab and found that anti-TGFβ Ab (10–20 μg/ml) abrogated the suppression induced by tolerized CD8+ T cells (data not shown). This is consistent with earlier data showing that contact-independent secretion of TGFβ is a major mechanism by which pCons-reactive CD8+ Ti cells suppress their targets (19).
CD8+CD28− T cells from pCons-tolerized BWF1 mice are weakly cytotoxic against B cells from nephritic BWF1 donors
We also examined the cytotoxicity of tolerized CD8+ T cells and CD8+CD28− T cells against B cells from BWF1 mice of various ages. We found that CD8+ T cells from mice treated with pCons peptide could stimulate slight but statistically significant cytotoxicity against B cells from nephritic animals (expressing anti-DNA surface B cell receptors), as shown in the open symbols in Fig. 6,A (which shows data for CD8+CD28− T cells). If the CD8+CD28− T cells were obtained from BWF1 mice tolerized 1 wk earlier with pCons, the cytotoxic capacity of the cells was low without stimulation, and it increased significantly by stimulation with pCons or pNeg (Fig. 6 A). At the 50:1 E:T ratio, pCons activated cytotoxicity against B cells significantly more than pNeg, but that difference was lost at the 100:1 E:T ratio. If the B cells were obtained from BWF1 females before the appearance of anti-DNA in sera, they were not subject to the increase in cytotoxicity from CD8+ T cells of tolerized mice (data not shown).
To measure granzyme, an effector molecule of cytotoxicity, we performed FACS analysis of quantities and percentage expression of granzyme B in CD8+ T cells from tolerized mice compared with naive mice. MFI for granzyme B was increased in CD8+ T cells from tolerized mice (Fig. 6,B, left panel). From these data, we conclude that pCons tolerization induces slightly increased cytotoxicity by CD8+CD28− T cells, but there is little specificity for the tolerizing peptide; the target of cytotoxicity is B cells from anti-DNA-positive mice. This contrasts with the effects of pCons tolerization on suppressing proliferation of syngeneic CD4+ T and B cells from either young or anti-DNA-positive mice via contact-independent secretion of TGFβ (Fig. 4, A and B and Ref. (19)), suggesting that tolerized CD8+ T cells contain subsets with different mechanisms of suppression, with the major one involving inhibition of both CD4+ T cells and B cells via secretion of TGFβ, and a possible minor effect involving cytotoxicity.
pCons treatment renders CD8+ Ti cells resistant to apoptosis
Because apoptosis has been linked to tolerance/anergy, we investigated apoptosis in our animal model of SLE after tolerization with pCons. Apoptosis assays with annexin V staining were performed on splenocytes 1 wk after tolerization, and stained after 2–8 h. As shown in Fig. 6,C–E, pCons-treated CD8+ Ti cells showed significantly less annexin V staining than did naive CD8+T cells. We used different positive and negative control peptides with pCons to test peptide specificity in altered apoptosis, because CD8+ T cells that suppress via secretion of TGFβ are activated only by their specific peptide (18). Negative control peptide-treated CD8+ T cells had more apoptosis than did those treated with pCons (Fig. 6,C–E). To test whether CD8+ Ti cells are relatively resistant to apoptosis, we induced apoptosis in naive and CD8+ Ti cells by anti-Fas ligand (FasL) Ab treatment and then measured the annexin V and 7AAD staining. The data shown in Fig. 6 F indicate that pCons-treated CD8+ Ti cells are indeed resistant to apoptosis.
pCons treatment changes the expression of PD1
To learn whether pCons treatment induced changes in the cell-surface expression of PD1, we performed flow cytometry for cell-surface expression of PD1 on CD8+ T cells by FACS. We found that PD1 expression and related MFI were reduced in all experiments comparing tolerized to naive CD8+ T cells (Fig. 7). Our time kinetics experiments with PD1 showed that PD1 expression was reduced as early as 1 wk after tolerization and persisted for 8 wk (Fig. 7,A–D). Both the percentage expression (Fig. 7,E) and MFI of PD1 were reduced significantly (Fig. 7,F, p < 0.0018). mRNA of PD1 was also down-regulated up to 6 wk after tolerization in CD8+Tcells (Fig. 7,G–I). Furthermore, time kinetics experiments indicated that tolerized CD8+ T cells suppressed proliferation of naive CD4 and B cells at the times PD1 expression was down-regulated. Suppression capacity was sustained for 6 wk, similar to PD1 down-regulation (Fig. 4 C). These data suggest that one of the mechanisms participating in immune tolerance in our model may be mediated via PD1. In summary, the data in this group of experiments indicate that pCons treatment increases intracellular Foxp3 expression and decreases PD1 surface expression in CD8+ T cells from tolerized mice.
Blocking of PD1/PDL1 interaction in naive and tolerized CD8+ T cells.
To address the effects of blocking of PD/PDL1 interactions in naive and tolerized CD8+ Ti cells from pCons-treated mice, we treated isolated cells with PD1- or PDL1-neutralizing mAbs in vitro to see the effects on proliferation of CD8+ T cells. We found that PD1 blocking in naive CD8+ T cells increases Foxp3 expression 3–4-fold (Fig. 7,J). However, blocking of PD1 in tolerized CD8+ Ti cells decreases Foxp3 expression (Fig. 7,K). Blocking of PD1/PDL1 abrogates the suppression of proliferation induced by tolerized CD8+ T cells (Fig. 7 L, compare column b to c (p < 0.0173) and column b to d (p < 0.05)). These data substantiate the relation of PD1 expression and PD1/PDL1 interactions in the suppressive capacity of CD8+ T cells in our model of immune tolerance, and they suggest that Foxp3 and PD1 regulation are related.
In the present study, we show that 1 wk after a single injection of pCons, CD8+ T cells are expanded more than other peripheral cell subsets, and they acquire the ability to directly suppress both naive CD4+ T cells and naive B cells, primarily by contact-independent cytokine secretion. Recent evidence suggests that CD8+ Ti cells suppress CD4+ T cell functions by rendering APCs “tolerogenic” (41, 42, 43). However, the studies reported herein show direct suppression of proliferation in vitro of both T and B cells in cultures in which APCs from naive mice were irradiated and should therefore only present peptides. Studies of the effect of our tolerization protocol on dendritic cells are in progress.
Several self-peptides important in SLE, in addition to pCons, have been shown to induce immune tolerance in lupus-prone mice; most of these work in part by induction of regulatory CD4+CD25+ and inhibitory CD8+ T cells. For example, Kang and colleagues showed that very low-dose tolerance with nucleosomal histone peptide could induce potent CD4+CD25+ regulatory T cells and CD8+ inhibitory T cells that control murine lupus (17). Adoptive transfer of Treg cells and CD8+ T cells resulted in suppression of IFN-γ secretion by nucleosomal-reactive CD4+ helper T cells and reduced autoantibody production (8, 9, 10, 17). Similarly, a peptide based on the CDR 1 of an autoantibody to DNA ameliorated murine lupus by up-regulating CD4+ T regulatory and CD8+ T cells; also, increased secretion of TGFβ by effector cells was found in that model (44, 45, 46).
In our model, several mechanisms of immune tolerance can be identified between 1 and 6 wk after a single tolerization. These include induction of anergy in CD4+ Th cells, induction of peptide-binding CD4+CD25+ regulatory T cells that suppress CD4+CD25− T cell proliferation by contact involving membrane-bound TGFβ and glucocorticoid-induced TNF receptor, and induction of CD8+ T cells that suppress autoimmunity by secretion of TGFβ (5, 6, 18). The work reported herein adds to this information that: 1) both CD8+CD28+ and CD8+CD28− inhibitory subsets express Foxp3, and silencing of Foxp3 in each subset abrogates inhibition; 2) the inhibitory CD8+ T cells are resistant to apoptosis; 3) the CD8+ inhibitory T cells can prevent proliferation in both CD4+ T cells and B cells; 4) CD8+ T cells can be cytotoxic for B cells from anti-DNA-positive mice; 5) suppression of surface PD1 expression on CD8+ T cells correlates with their increased expression of Foxp3 and their ability to suppress anti-DNA production; and 6) silencing Foxp3 or inhibiting the activity of PD1/PDL1 interactions are equally effective in abrogating the suppressive capacity of CD8+ T cells from tolerized mice.
We show for the first time in this system a small but statistically significant population of nonpeptide-specific CD8+ CTLs in tolerized mice that can kill B cells from anti-DNA-positive, but not anti-DNA-negative, mice, suggesting that the B cells must express anti-DNA Abs (probably with peptides similar to pCons in the VH regions) to be subject to nonspecific killing. These data are consistent with the work reported by Fan and Singh (47) showing that vaccination of young BWF1 mice with plasmids encoding MHC class I determinants induced cytotoxic CD8+ T cells that killed B cells from old, but not young, BWF1 females. Such CD8+ T cells were powerful enough to suppress anti-DNA production in vivo and prolong survival.
Our time kinetics experiment regarding granzyme B expression showed that after tolerization with pCons, granzyme B expression was increased for 3–7 days and then gradually returned to baseline (data not shown). This is in contrast to the induction of peptide-specific CD4+CD25+ Treg and peptide-specific CD8+ Ti cells that lasts up to 8 wk after a single dose of pCons (18). This provides evidence that the role of cytotoxic killing of B cells in this tolerance system is minor, compared with the durability and power of both the CD4+CD25+ Treg and the CD8+ Ti cells. It is these latter two types of cells that probably have a sustained effect on both CD4+ helper T cells and B cells that result in clinical benefit. In contrast, it is possible that new waves of cytotoxic CD8+ T cells could be generated under permissive conditions. In this regard, after peptide vaccination of melanoma patients, a distinct CD8+CD28−, granzyme-positive subset appeared. In the presence of IL-12, a new subset of cytotoxic CD8+ T cells could be generated from the CD8+CD28+, low-granzyme subset (48).
The importance of reduced expression of PD1 in the CD8+ T cells from tolerized mice is not clear. PD1 (a member of the CD8 family), bound by one of its ligands, PDL1, generally provides negative signals to T cells (including CD8+ T cells), resulting in cell anergy and decreased capacity to generate cytotoxicity (49, 50, 51, 52). Tissue expression of PDL1 mediates peripheral T cell tolerance, probably by binding to PD1 on potentially harmful cells that have homed to that tissue. Although sustained expression of PD1 on CD8+ T cells is an indicator of cell exhaustion (51), it may be that down-regulation is required to release the inhibitory capacity of the cell. PD1 was elevated in CD8+ T cells in a hepatitis C viral infection model (where infection is associated with hepatitis C virus-specific CD8 exhaustion) (53). In a recent report of PD1/PDL1 participation in immune tolerance to a self-Ag, PD1 was up-regulated primarily during the first and second waves of CD8+ T cell proliferation (1–3 days after exposure to a high dose of self-Ag in mice), and its expression was declining by day 7 (54). It is likely that PD1 expression must be down-regulated to permit full generation of CD8+ Ti cells and cytotoxic CD8+ T cells, and we may have identified this down-regulated state at the 5 day–6 wk time points studied in our experiments. Furthermore, we found that PD1 blocking in naive CD8+ T cells increases Foxp3 expression 3–4-fold (Fig. 7,J). However, blocking PD1 in tolerized CD8+ Ti cells further decreases Foxp3 expression (Fig. 7,K). Also, blocking PD1/PDL1 abrogates the suppression of proliferation induced by tolerized CD8+ T cells (Fig. 7,L). siRNA silencing of Foxp3 reduced PD1 cell-surface expression and mRNA in tolerized CD8+ T cells (Fig. 7 M). Our data support a role for suppression of PD1/PDL1 in naive cells that results in increased Foxp3 expression in naive cells (probably rendering them suppressive). We imagine that down-regulation of PD1 is linked to acquisition of suppression, but complete abrogation in tolerized cells disables that suppression and would favor autoimmunity, consistent with the report that PD1 knockout mice develop SLE-like disease (55). Therefore, the PD1/PDL1 molecules are involved in tolerance in our model, and their balance tilts from tolerance toward autoimmunity. Future detailed studies are required to test this hypothesis.
In any case, note that in all these experiments, mice were tolerized at age 10–14 wk, before anti-DNA appears in serum, and before they develop the defects of T cells associated with aging in this strain (low IL-2 production by CD4+ T cells and defective proliferation of CD8+ T cells with activation resulting in Ag-induced death rather than expansion (4, 6, 7)).
Finally, our gene-silencing study showed the importance of Foxp3 in both CD8+ CD28+ and CD8+CD28− T cells in our model. Inhibition of Foxp3 abrogated the suppression induced by tolerized CD8+CD28− T cells. A siRNA-silencing study of Foxp3, but not of Trp53 or CCR7, abrogated the suppressive capacity of tolerized effectors, both CD4+CD25+ T cells and the whole population of CD8+ T cells (6, 18). We show that both CD8+ subsets identified by expression or lack of expression of CD28 participate in inhibition of anti-DNA and depend at least in part on expression of Foxp3. Those cells up-regulate mRNA for Foxp3 and TGFβ, but not IFN-γ or IL-10, consistent with earlier work showing the dependence of suppression by the CD8+ T cells on TGFβ secretion (6). Of note, another group studying CD8+ T cells from a model of immune tolerance showed that CD8+Foxp3+ suppressor T cells mediate immune tolerance to allogeneic heart transplants by inducing Foxp3, PIRβ, and Ig-like transcript 3 and 4, thereby inhibiting alloreactivity (41, 56). The molecular mechanisms that govern the suppression of disease and the role of these genes in the CD8 suppressors are not completely clear. A future study will require detailed functional analysis of molecular mechanisms and selected genes in autoimmunity.
In conclusion, the effects of tolerization of young BWF1 females with a single dose of pCons include expansion of CD8+ T cells that can suppress autoimmunity via previously unidentified mechanisms. These mechanisms now point to possible new target molecules in selected cell subsets with translational and clinical importance in this model of induced immune tolerance.
We thank the University of California Los Angeles Flow Cytometry Core Facility for FACS analysis and cell sorting, Dr. Desmond Smith laboratory at University of California–Los Angeles for the real-time PCR assay, Fanny Ebling for pCons tolerization, and Robert J. Rooney, Sven De Vos, and David Elashoff for microarray and statistical analyses.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health Grants AI46776 (to B.H.H.), AR53239 (to A.L.C.), AI 065645, and AR054034 (to R.P.S.), University of California–Los Angeles Senate Core Grant (to R.P.S. and B.H.H.), Arthritis Foundation Southern California Chapter Grant (to B.H.H. and R.P.S.), and gifts from the Maltz Laboratory, the Horchow family, and Jeanne Rappaport.
Abbreviations used in this paper: SLE, systemic lupus erythematosus; 7AAD, 7-aminoactinomycin D; FasL, Fas ligand; FSC, forward scattering cells; MFI, mean fluorescence intensity; pCons, pConsensus; pNeg, negative control peptide; siRNA, small interfering RNA; Ti, inhibitory T; Treg, regulatory T.