Nature Immunology5, 516 - 523 (2004)
Published online: 18 April 2004; | doi:10.1038/ni1063
Analysis of regulatory CD8 T cells in Qa-1-deficient mice
Dan Hu1, 2, Koichi Ikizawa1, 2, Linrong Lu1, Marie E Sanchirico1, Mari L Shinohara1
& Harvey Cantor1
1
Department of Cancer Immunology and AIDS, Dana-Farber Cancer Institute, Department of Pathology, Harvard Medical School, Boston, Massachusetts
02115, USA.
The mouse protein Qa-1 (HLA-E in humans) is essential for immunological protection and immune regulation. Although Qa-1 has been linked to CD8 T cell−dependent suppression, the physiological relevance of this observation is unclear. We generated mice deficient in Qa-1 to develop an understanding of this process. Qa-1-deficient mice develop exaggerated secondary CD4 responses to foreign and self peptides. Enhanced responses to proteolipid protein self peptide were associated with resistance of Qa-1-deficient CD4 T cells to Qa-1-restricted CD8 T suppressor activity and increased susceptibility to experimental autoimmune encephalomyelitis. These findings delineate a Qa-1-dependent T cell−T cell inhibitory interaction that prevents the pathogenic expansion of autoreactive CD4 T cell populations and consequent autoimmune disease.
Successful progression of thymocytes through positive selection requires expression of T cell receptors (TCRs) capable of efficient binding to self major histocompatibility complex (MHC) products. Although T cell clones bearing TCRs with high affinity for self peptide−MHC products are generally eliminated in the thymus, the TCR repertoire of mature T cells is strongly biased toward self reactivity1. As a result, substantial numbers of mature T cell clones can specifically proliferate in response to self peptide−MHC complexes, and some can differentiate into effector cells in the context of inflammatory stimuli2,
3,
4.
Expansion of autoreactive T cell populations is constrained in part by abortive TCR signals that lead to T cell elimination or inactivation5,
6. However, functional elimination of autoreactive T cells through activation-induced cell death (AICD) and anergy does not prevent expansion and consequent autoimmune disease after deliberate immunization7 or viral infection8,
9. These findings have stimulated research into T cells and other cell types that may exert dominant inhibitory effects on expansion of pathogenic autoreactive T cell populations.
A regulatory CD4+ T cell sublineage that often expresses CD25 and the Foxp3 transcription factor can inhibit expansion of autoreactive T cell populations in lymphopenic inflammatory environments9,
10,
11. Other studies have suggested that a subpopulation of CD8 T cells might suppress the response of successfully activated CD4 T cells and B cells through an interaction that depends on expression of the MHC class Ib molecule Qa-1 on activated target cells12,
13. Interest in this area has been rekindled by findings that an inhibitory interaction between CD8 T cells and autoreactive CD4 T cells may be required for immunological resistance to experimental autoimmune encephalomyelitis (EAE), a mouse model of multiple sclerosis14,
15. These studies have reopened the possibility that a subpopulation of CD8 T cells may mediate physiologically important inhibitory effects on autoimmune responses through TCR-dependent recognition of Qa-1 on activated lymphocyte target cells.
Qa-1, the mouse homolog of human leukocyte antigen E (HLA-E), forms a heterodimer with 2-microglobulin that binds to and presents peptides derived from self or foreign proteins16 after deliberate immunization or infection17,
18,
19. Although the mRNA encoding Qa-1 is detectable in many cell types20, surface expression of the Qa-1−2-microglobulin heterodimer may be highly constrained by Qa-1−2-microglobulin assembly and/or transport mechanisms, which favor expression in activated T and B lymphocytes and dendritic cells19,
21.
Peptide-containing Qa-1 complexes engage two broad classes of receptors. Qa-1 heterodimers containing peptides derived in a TAP (transporter associated with antigen processing)−dependent way from MHC class Ia leader sequences, called Qdm (for Qa-1 determinant modifier), bind to nonclonally distributed CD94-NKG2A, CD94-NKG2C and/or CD94-NKG2E receptors on natural killer (NK) cells and a subpopulation of CD8 T cells. The functional outcome of Qdm-NKG2A interactions is generally inhibition of NK activity or CD8 cytotoxic T lymphocyte (CTL) activity, as judged from analysis of antiviral responses22. A second class of Qa-1 ligand comprises Qa-1−2-microglobulin heterodimers that contain peptides of a growing list, including those derived from TCR V, preproinsulin, bacterial GroEL and HSP60 (refs. 18,23,
24,
25,
26). Interaction between this set of Qa-1 ligands and the TCR on CD8 T cells can promote CD8 T cell activation, population expansion and expression of effector cell activity.
We generated mice deficient in Qa-1 expression to develop a clearer understanding of the potential regulatory activity of Qa-1-reactive CD8 T cells. Qa-1-deficient mice developed exaggerated secondary CD4 responses after viral infection or immunization with foreign and self peptides. Enhanced responses of Qa-1-deficient mice to proteolipid protein (PLP) self peptide were associated with failure to develop the EAE resistance that developed after immunization of Qa-1 wild-type mice with PLP in the absence of pertussis toxin. Failure to develop resistance to EAE was associated with escape of Qa-1-deficient CD4 T cells from CD8 T cell suppression, which was restored by lentiviral-based expression of the syngeneic allele encoding Qa-1. Because Qa-1-restricted CD8 T cells selectively inhibited the pathogenic portion of the CD4 T cell anti-peptide response, this inhibitory CD8-CD4 T cell interaction may prevent expansion of pathogenic autoreactive CD4 T cell populations and consequent autoimmune disease.
Results Generation of Qa-1-deficient mice CD8 T cells that recognize antigen in association with Qa-1 (HLA-E in humans) have been linked to immunoregulation13. Insight into the potential physiological function of Qa-1-restricted CD8 T cells depends in part on the generation and analysis of mice deficient in Qa-1 expression secondary to targeted gene mutation. Production and definition of these mice through targeted mutation has been complicated by the presence of a series of highly homologous genes, including other MHC class I genes and an almost identical pseudogene (T10c), adjacent to the gene encoding Qa-1 that can influence recombination events and confound analysis of recombined DNA.
We screened a mouse bacterial artificial chromosome library by Southern blot analysis with a probe derived from the third exon of the gene encoding Qa-1. A fragment 11.5 kilobases (kb) in length from a clone containing the gene encoding Qa-1 was unambiguously identified according to PCR analysis and partial DNA sequencing followed by complete DNA sequencing. After being subcloned into pPCR-Script Amp SK(+), the mapped and fully sequenced genomic DNA fragment was used to generate a replacement vector with positive (neomycin resistance) and negative (diphtheria toxin A chain) selection markers. Recombination resulted in deletion of exons 1, 2 and 3 of the gene encoding Qa-1 and introduction of a frameshift predicted to completely abrogate Qa-1b expression (Fig. 1a). We transfected mouse embryonic stem cells with the linearized vector and screened the cells by PCR and Southern blot analysis for the recombinant allele, then used selected embryonic stem cell clones (Fig. 1b, middle) for microinjection into blastocysts. We screened the resultant chimeras and their offspring for their Qa-1 genotypes by PCR (Fig. 1b, left) and Southern blot analysis (Fig. 1b, right). We activated thymocytes and splenic T cells from Qa-1-deficient mice with concanavalin A (ConA); the absence of Qa-1b expression on these cells, as determined by flow cytometry with monoclonal antibody to Qa-1b, confirmed the Qa-1-deficient phenotype at the protein level (Fig. 1c). We then backcrossed Qa-1-deficient (C57BL/6 129) mice four generations to C57BL/6 mice for all of the following experimental analyses except those involving herpes stromal keratitis (HSK), for which we compared C57BL/6 129 Qa-1-deficient and wild-type littermates. Here, lymphoid cells from C57BL/6 Qa-1-deficient (F4 generation) mice are called 'B6 Qa-1-deficient', and C57BL/6 Qa-1 wild-type or heterozygous (F4) littermates are called 'B6 Qa-1 wild-type'. B6 Qa-1-deficient and B6 Qa-1 wild-type mice at 4−6 weeks of age did not show gross abnormalities and the proportions of thymocyte and T cell subpopulations in B6 Qa-1 wild-type and B6 Qa-1-deficient mice were not distinguishable (Supplementary Table 1 online). We also irradiated hosts (900 rads) deficient in recombination activating gene 2 (RAG-2) or RAG-2−perforin, reconstituted them with 1:1 mixtures of B6 Qa-1-deficient and B6 Qa-1 wild-type bone marrow and analyzed T cell development in these mice to determine whether Qa-1 deficiency affects the numbers or phenotype of developing thymocytes secondary to enhanced perforin-dependent NK activity27. Twelve weeks after reconstitution, the proportions of B6 Qa-1-deficient and B6 Qa-1 wild-type CD4 or CD8 T cells (distinguished by Thy-1.1 and Thy-1.2 markers) in thymus, lymph node and spleen were similar in RAG-2-deficient and RAG-2−perforin-deficient mice (Supplementary Fig. 1 online). Nevertheless, the experiments of adoptive or adaptive immune responses of Qa-1-deficient and Qa-1 wild-type T cells described below used RAG-2−perforin-deficient hosts that lacked detectable NK activity.
Figure 1. Identification of targeted Qa-1 gene by PCR and Southern blot.
(a) Qa-1 genomic locus and targeting strategy. H, HindIII; X, XbaI; B, BamHI; E, EcoRV; Hp, HpaI; DT, diphtheria toxin; neor, neomycin-resistance gene; WT, wild-type; KO, Qa-1-deficient. (b) Left, PCR of mouse genomic DNA; the 1,340-bp fragment corresponds to the mutant allele, and the 757 bp fragment corresponds to the wild-type allele. M, 100-bp DNA ladder (sizes, left margin). Middle, Southern blot of embryonic stem cell genomic DNA. Right, Southern blot of mouse (B6 129) genomic DNA. Upper bands (8.1 kb) correspond to the mutant allele; lower bands (7.1 kb) correspond to the wild-type allele. +/+, B6 Qa-1 wild-type; +/-, B6 Qa-1 heterozygous; -/-, B6 Qa-1-deficient. (c) ConA-induced Qa-1 expression. Splenocytes and thymocytes from littermates (four Qa-1 wild-type (+/+), five heterozygous (+/-) and five Qa-1-deficient (-/-)) were individually stimulated with ConA for 40 h and analyzed with anti-Qa-1b. Data represent mean numbers of Qa-1+ cells and standard errors.
Response to herpes simplex virus 1 Ocular herpes simplex virus 1 (HSV-1 (KOS)) infection induces a vigorous CD4-dependent autoimmune response in the corneas of susceptible mouse strains (such as BALB/c and C.AL-20) but not resistant strains, including C57BL/6 and 129, which is mediated at least in part by CD4 T cells that recognize self peptides that cross-react with HSV-derived peptides8. HSV-1 (KOS)−inoculated Qa-1-deficient animals developed HSK (disease index, 19.5), whereas wild-type littermates did not (Fig. 2a). Unseparated lymph node cells as well as NK1.1+-depleted CD4 T cell populations (NK1.1-) and NK1.1+-depleted and CD25+-depleted CD4 T cell populations (NK1.1-CD4+CD25-) from HSV-1-infected Qa-1-deficient donors showed enhanced interferon- (IFN-) responses compared with their counterparts from HSV-1-infected wild-type mice after restimulation in vitro by virus or viral protein (glycoprotein D; Fig. 2b,c). These findings suggested that the enhanced HSK response of Qa-1-deficient mice was associated with upregulated IFN- production by CD4 T cells.
Figure 2. The response of Qa-1-deficient mice to HSV-1 (KOS) infection.
(a) Left, HSK induction in Qa-1-deficient () and Qa-1 wild-type () mice infected in the right eye with HSV-1; disease was assessed on days 9 and 12 (refs. 40,41). Percentage of each group with detectable disease (incidence) and disease severity are based on analysis of HSK in four mice per data point. Representative of three independent experiments. Right, targeted disruption of Qa-1 converts HSK susceptibility. B6 129 Qa-1-deficient (KO) or Qa-1 wild-type (WT) mice were infected with HSV-1 KOS as described40. Disease index = percent incidence mean severity of clinical stromal keratitis 10. Data represent mean of 20 (WT) and 29 (KO) mice 2 weeks after infection. (b) Draining lymph nodes (3 105 cells/well) 16 d after HSV-1 (KOS) infection were stimulated with glycoprotein D (50 g/ml) plus irradiated wild-type splenocytes (3 105 cells/well) as antigen-presenting cells. IFN- production was measured after 3 d by ELISA. Data represent two independent experiments (Expt I and Expt II). (c) Purified (NK1.1-) CD4 T cells or CD4 T cell populations depleted of CD25+ (CD4+CD25-) cells from B6 Qa-1-deficient and B6 Qa-1 wild-type mice were stimulated with glycoprotein D (GlyD; 50 g/ml), HSV-1 (KOS; 5 105 plaque-forming units/well, inactivated) or anti-CD3 (5 g/ml). Supernatants were collected on day 3 and IFN- production was measured by ELISA.
Qa-1-restricted suppression of the V3+ CD4 T cell response V3+ CD4 T cells have a high affinity for superantigen staphylococcal enterotoxin A (SEA) and, after activation, these cell populations initially expand before a considerable reduction in cell numbers occurs because of AICD28. Here, the numbers of V3+ CD4 T cells after two SEA injections 2 weeks apart decreased by 60−70% in wild-type mice but not in Qa-1-deficient mice (data not shown). To define the potential involvement of Qa-1-dependent inhibition by CD8 T cells, we adoptively transferred CD8 T cells from SEA-immune B6 Qa-1b or B6 Tlaa (Qa-1a) donors along with CD4 T cells from B6 Qa-1b or B6 Tlaa (Qa-1a) mice into adoptive B6 RAG-2−perforin-deficient recipients. CD8 T cells from SEA-primed B6 Qa-1b mice inhibited about 75% of SEA-induced expansion of B6 Qa-1b V3+ CD4 T cell populations but did not inhibit superantigen-induced expansion of B6 Qa-1-deficient or B6 Qa-1a V3+ CD4 T cell populations. Conversely, CD8 T cells from SEA-primed B6 Tlaa (Qa-1a) donors inhibited the adoptive anti-SEA response of B6 Tlaa (Qa-1a) but not B6 Qa-1b CD4 T cells (Fig. 3). Although the response of mixtures of CD8 and CD4 T cells that expressed differing Qa-1 alleles was occasionally enhanced compared with that of CD4 T cells alone (data not shown), this was not a consistent finding. The observation that the inhibitory activity of (NK1.1-) CD8 T cells is restricted by Qa-1 alleles is consistent with Qa-1-dependent peptide presentation rather than engagement by nonclonally distributed NKG2 receptor(s), which would not be expected to discriminate between Qa-1 alleles. These findings suggest an inhibitory T cell−T cell interaction in which CD8 T cells recognize a Qa-1-restricted target on V3+ CD4 T cells.
Figure 3. Induction of suppressive CD8 activity by SEA.
CD8 T cells from SEA-immune B6 and B6 Tlaa mice were transferred into RAG-2−perforin-deficient mice (1 106 cells/mouse) along with nonimmune CD4 T cells (1 106 cells/mouse), and recipient mice were boosted with SEA (5 g, intraperitoneally). V3+ CD4 T cells in lymph nodes and spleen were counted 5 d later (three to four mice per group).
Susceptibility to EAE We next examined the development of EAE after immunization of B6 Qa-1 wild-type and B6 Qa-1-deficient mice with PLP peptide (amino acids 172−183), an MHC class II binding peptide, in complete Freund's adjuvant (CFA; called 'PLP-CFA' here; pertussis toxin challenge 2 d later). Although primary EAE began earlier in the B6 Qa-1-deficient than B6 Qa-1 wild-type mice after primary immunization (mean disease onset, day 11 versus day 21), disease incidence and intensity was similar over the 6-week observation period in mutant and wild-type mice (Fig. 4a).
Figure 4. Protective effects of peptide preimmunization on the development of EAE.
(a) Clinical score (right) and incidence (left) of EAE in B6 Qa-1b wild-type (WT) and B6 Qa-1-deficient (KO) mice immunized with PLP peptide (150 g) in CFA plus pertussis toxin on day 0 and boosted with pertussis toxin on day 2 (four to six mice per group). (b) Effects of preimmunization with PLP peptide. B6 Qa-1b wild-type (WT) and B6 Qa-1-deficient (KO) mice were preimmunized with PLP peptide (10 g) in CFA and were boosted 10 d later with PLP peptide (150 g) in CFA plus pertussis toxin (PLP-(C/P)). Clinical score (right) and incidence (left) were determined after the boost (time, horizontal axes; eight to ten mice per group). *, P = 0.04. (c) Effects of preimmunization with PLP peptide. B6 Qa-1b wild-type mice were preimmunized with PLP peptide (10 g) or heat shock protein peptide (100 g; GMKFDRGYI) in CFA (CFA-PLP and CFA (hsp), respectively) or with CFA alone (CFA) or were left unimmunized (None) and were boosted 10 d later as described in b. Mice were monitored for 40 d to determine disease incidence and maximum score (five to six mice per group).
Immunization with myelin basic protein (MBP) leads to an episode of EAE in susceptible strains (such as PL/J) that is often followed by remission and CD8 T cell−dependent resistance14,
29. The C57BL/6 mouse strain is relatively resistant to EAE after immunization with PLP-CFA and requires injection of pertussis toxin (twice) for development of substantial disease30,
31,
32. Because immunization of B6 mice with PLP-CFA without pertussis toxin is sufficient to induce a strong anti-PLP CD4 T cell response but no disease, we sought to determine whether this form of immunization might also induce CD8 T cell−dependent resistance. Although immunization of B6 mice with PLP-CFA without pertussis toxin did not induce substantial clinical disease over the next 28 d, PLP-CFA-immune B6-Qa-1 wild-type mice were almost completely resistant to EAE development after subsequent immunizations with PLP-CFA plus pertussis toxin (Fig. 4b); identical PLP-CFA preimmunization of B6 Qa-1-deficient mice had almost no protective effect on subsequent EAE incidence or intensity after secondary and tertiary challenges with PLP-CFA plus pertussis toxin. These results suggest that EAE resistance induced by immunization with PLP-CFA without pertussis toxin is Qa-1 dependent.
We then sought to determine whether the protective effects of PLP-CFA exposure were peptide specific, as nonspecific activation of NK and NKT cells generally before (but not after) exposure to myelin-derived peptide can potentiate subsequent EAE33. EAE resistance depended on specific immunization with PLP, as we did not find resistance after immunization with CFA alone or CFA plus an irrelevant peptide (heat shock protein) in B6 Qa-1 wild-type mice (Fig. 4c). Initial exposure to PLP without pertussis toxin conferred Qa-1-dependent protection against EAE development and was immunologically specific.
Qa-1-restricted inhibition of CD4 responses to PLP Reduced secondary responses of CD4 T cells to superantigen was associated with Qa-1-restricted inhibition by CD8 T cells. We sought to determine whether a similar inhibitory CD8-CD4 interaction was associated with Qa-1-dependent resistance to EAE after PLP immunization. Expansion of CD4 T cell populations from B6 Qa-1b PLP-CFA-immune donors in adoptive hosts challenged with PLP was inhibited by CD8 T cells from B6 Qa-1b PLP-immune donors. As noted from analysis of the adoptive SEA response, susceptibility of CD4 T cells to CD8 T cell−dependent inhibition required expression of the allele encoding Qa-1 expressed during the primary response (data not shown): both Qa-1-deficient CD4 T cells and B6 Qa-1a CD4 T cells were resistant to the inhibitory effects of Qa-1b-expressing CD8 T cells (Fig. 5a). Thus, suppression of autoreactive CD4 T cells by CD8 T cells is not only Qa-1 dependent but also allele specific.
(a) B6 Qa-1 wild-type (WT) and B6 Qa-1-deficient (KO) mice were immunized with PLP peptide (150 g) in CFA plus pertussis toxin, then CD4 T cells (1 106) and CD8 T cells (2 106) were collected for adoptive transfer into RAG-2−perforin-deficient mice 10 d later along with subcutaneous injection of PLP peptide (20 g) in CFA. CD4 T cell numbers in draining lymph nodes and spleen were determined 12 d later (three mice per group). (b) RAG-2−perforin-deficient hosts of different T cell subsets (left margin) were immunized with PLP peptide (150 g) in CFA plus pertussis toxin and were assigned scores for EAE for 60 d. After this monitoring, draining lymph node cells were stimulated with PLP peptide (10 g/ml) and interleukin 2 and IFN- production was measured after 72 h. The response of recognition of CD4 T cells alone is considered 100% in this comparison. (c) Purified CD4 T cells were infected and, after being washed, 2 106 cells were stimulated in vitro with plate-bound anti-CD3 (1 g/ml) and anti-CD28 (5 g/ml), and surface expression of Qa-1b was determined by immunofluorescence 48 h later with biotinylated anti-Qa1b and phycoerythrin-conjugated streptavidin. Activated GFP-infected Qa-1-deficient and Qa-1b wild-type CD4 T cells represent negative and positive controls, respectively. Percentages (%) represent Qa-1+ cells. (d) Lentiviral expression of the Qa-1b molecule restores sensitivity of B6 Qa-1-deficient CD4 T cells to CD8 suppression. Donor CD4 and CD8 T cells were prepared as described in a. B6 Qa-1-deficient donor CD4 T cells (1106) were infected with Qa-1b lentiviral vector (as described in c) before transfer into RAG-2−perforin-deficient mice alone or with CD8 T cells (2 106) before immunization with PLP peptide (150 g) in CFA, and were killed 40 d later (three mice per group).
To investigate the relationship between the Qa-1-restricted CD8-CD4 interaction, we measured the development of CD4 T cell responses to PLP in adoptive RAG-2−perforin-deficient hosts infused with CD4 T cells and CD8 T cells from PLP-CFA-immune donors. The inhibitory effects of CD8 T cells on PLP-dependent CD4 population expansion, described above, was associated with reduced development of EAE. Co-transfer of CD8 T cells from either PLP-CFA-immune B6 Qa-1b or B6 B6-Tlaa F1 (Qa-1a/b) CD8 T cells along with B6 Qa-1b CD4 T cells inhibited development of EAE, whereas co-transferred CD8 T cells from B6 Qa-1a donors did not. Reduced EAE development in adoptive hosts was associated with a substantial reduction in CD4-dependent interleukin 2 and IFN- cytokine responses after in vitro restimulation with PLP peptide (Fig. 5b). Resistance of B6 Qa-1-deficient CD4 T cells to the inhibitory effects of CD8 T cells reflected loss of a targeting molecule rather than changes in CD4 T cells secondary to altered development in Qa-1-deficient mice, as lentivirus-dependent expression of Qa-1b in Qa-1-deficient CD4 T cells restored susceptibility of these cells to CD8-dependent inhibition (Fig. 5c,d).
CD8 T cell inhibition of the OT-2 TCR response We concluded that the Qa-1 suppressive interaction between CD8 and CD4 T cells is mediated by the CD8 TCR rather than nonclonally distributed NKG2-type receptors, because CD8-dependent suppression was restricted to CD4 T cells that expressed the allele encoding Qa-1 present during immunization. According to this view, CD8 T cells that express a single and irrelevant TCR should be unable to mediate suppressive activity. We tested this by examining adoptive anti-peptide responses of CD4 T cells that bear the OT-2 TCR transgene that recognizes an ovalbumin (OVA) peptide. After OT-2 T cell vaccination of B6 mice, we purified CD8 T cells from OT-2-immunized mice and transferred the cells (2 106) along with OT-2 V5+ CD4 T cells (1 106) into RAG-2−perforin-deficient mice before challenge with OVA peptide (50 g) in incomplete Freund's adjuvant (IFA). We collected draining lymph node cells 7 d later and counted OT-2 V5+ CD4 T cells, as V5 expression normally characterizes a greater population of CD8+ than CD4+ peripheral T cells, which is thought to be due to more efficient intrathymic positive selection on MHC class I (rather than class II) antigens.
Inclusion of CD8 T cells inhibited 70−80% of peptide-dependent expansion of Qa-1 wild-type OT-2 V5+ CD4 T cell populations, but did not inhibit responses of Qa-1-deficient OT-2 CD4 T cells (Fig. 6a). We sought to determine whether constraints placed on TCR expression might preempt CD8-dependent suppressive activity. CD8 T cells that expressed a single OT-1 TCR derived from RAG-deficient mice were unable to suppress the response of Qa-1 wild-type CD4 T cells bearing the OT-2 TCR transgene, despite activation with OT-1 peptide (Fig. 6b). We then investigated the CD8 T cell−dependent inhibitory effects on peptide-specific OT-2 CD4 T cells in vitro, after in vivo boosting in RAG-2−perforin-deficient adoptive hosts. We obtained lymph node and spleen cells from the adoptive hosts described above and stimulated these in vitro with 'graded' doses of OT-2 peptide. The continued presence of CD8 T cells during a 72-hour in vitro restimulation by OT-2 peptide notably inhibited peptide-induced expansion of Qa-1+ but not Qa-1- V5+ CD4 T cell populations. Although removal of CD8 T cells before in vitro restimulation partially restored V5+ cell population expansion, the IFN- response of these V5+ CD4 T cells to OT-2 peptide remained impaired (Fig. 6c). These data suggest that an activated, nonspecific, monoclonal CD8 T cell subpopulation may not be sufficient to mediate the suppressive activity observed after stimulation of the immune system.
Figure 6. T cell vaccination with OT-2 CD4 T cells induces suppressive CD8.
(a) After OT-2 T cell vaccination of C57BL/6 mice, purified CD8 T cells (2 106) from OT-2 immunized mice were transferred with OT-2 CD4 T cells (1 106) into RAG-2−perforin-deficient mice before challenge with OVA peptide (50 g) in IFA. Then, 7 d later, draining lymph node cells were collected and OT-2 CD4 T cells were counted (three mice per group). (b) Dose-response analysis of CD8-dependent suppression in adoptive RAG-2−perforin-deficient hosts. OT-2 CD4 T cells (1 106) were transferred into RAG-2−perforin-deficient hosts with CD8 T cells (numbers, horizontal axis) from OT-2-immunized C57BL/6 mice () or OT-1 TCR transgenic RAG-2-deficient C57BL/6 mice () before challenge with OT-2 peptide and counting of OT-2 cells 12 d later. Each data point represents mean s.d. of three mice per group. Co-transfer of CD8 T cells from B6 or OT-1 RAG-2-deficient donors did not result in notable inhibition of the response of Qa-1-deficient OT-2 CD4 T cells: CD4 alone, 8.8 0.2 106 cells versus B6 CD8 7.1 0.4 106 cells; OT-1 CD8, 8.4 0.7 106 cells. Even after immunization with OT-2 CD4 T cells, OT-1 CD8 T cells failed to mediate suppressive activity (data not shown). (c) OT-2 CD4 T cells (1 106) were transferred into RAG-2−perforin-deficient mice without (CD4) or with (CD4-CD8) OT-2-immune CD8 T cells (1 106) and recipient mice were challenged with OVA peptide (50 g) in IFA. After 7 d, the mixture of draining lymph node and spleen cells were prepared for in vitro restimulation (4 105 cells/well) with OVA peptide (concentration, horizontal axis). Cells from the CD4 and CD8 recipients were stimulated with OT-2 peptide (50 g) without CD8 depletion (CD4 plus CD8 T cells) or after CD8 depletion (CD4 T cells). After 72 h, V5+ CD4 T cells in each well were counted (left) and IFN- in the supernatant was measured by ELISA (right). Data represent the mean s.d. of triplicate cultures of Qa-1 wild-type CD4 T cells. The CD8 T cells showed no significant effect on Qa-1-deficient CD4 T cells. Representative of two independent experiments with similar outcomes.
Discussion To define the potential regulatory function of Qa-1 in the immune system and the existence of a subpopulation of Qa-1-restricted suppressive CD8 T cells, we generated Qa-1-deficient mice through targeted gene disruption. Although interactions between Qa-1 and NKG2A are known to inhibit NK cell activity and CD8 CTL activity22, our analysis of nonimmune Qa-1-deficient mice did not show an obvious defect in T cell development, that is, secondary to increased susceptibility of Qa-1-deficient T or B lymphocytes to NK-dependent attack34. Moreover, in the absence of immunization or deliberate infection, activation markers on CD4 and CD8 T cells and thymocytes were indistinguishable in wild-type and Qa-1 deficient mice, as were primary CD4 T cell responses to different types of T cell ligand. The main difference in the immune response phenotype between the two strains became apparent from analysis of the secondary responses to foreign and self peptides and to bacterial superantigen. Qa-1-deficient mice developed enhanced CD4 T cell responses after infection or immunization, and we delineated the existence of a subpopulation of Qa-1-restricted suppressive CD8 T cells that downregulated CD4 T cell immunity.
The V3+ CD4 response to SEA was associated with the development of CD8 T cells that exerted Qa-1-restricted inhibition of adoptively transferred anti-SEA-specific V3+ CD4 T cells. Although the initial responses of B6 Qa-1-deficient or wild-type mice to SEA were not obviously different, the responses of wild-type mice to SEA rechallenge showed a 60−65% decrease in V3+ CD4 T cells, whereas the second anti-SEA response of Qa-1-deficient mice was marked by increased numbers of V3+ CD4 T cells that persisted throughout the next 10 d (data not shown). The observation that CD8 T cell−dependent inhibition was limited to CD4 T cells that expressed the alleles encoding Qa-1 present during priming of CD8 T cells was consistent with TCR-dependent recognition and discrimination of peptide−Qa-1 complexes of target CD4 T cells rather than engagement of Qa-1 by NKG2 receptors, which would not be expected to discriminate between alleles encoding Qa-1. The idea that TCR recognition contributes to Qa-1-restricted suppression by CD8 T cells was further supported by the finding that CD8 T cells expressing a TCR specific for an OVA peptide were unable to mediate suppressive activity.
These observations also suggested that Qa-1-restricted regulatory CD8 T cells routinely develop during primary responses and act during secondary responses. Analysis of the T cell response to the PLP self peptide allowed us to test this view in the context of both CD4 anti-peptide responses and the development of EAE. Because immunization with PLP peptide-CFA without pertussis toxin induces strong CD4 T cell anti-PLP responses but not disease, we could determine whether expansion of CD4 T cell populations by PLP peptide was sufficient to induce CD8-dependent resistance to EAE after secondary PLP challenge. This was the case. It has been difficult to establish the specificity of CD8 T cell−dependent resistance to MBP-induced resistance in PL/J mouse strains, because MBP peptide is required for both initial induction of monophasic disease and reinduction of EAE. The PLP-B6 model allowed us to address this; immunization of B6 Qa-1 wild-type mice with PLP-CFA, but not CFA alone or CFA mixed with an irrelevant peptide or protein, was required for resistance to PLP-induced EAE. Analysis of the cellular basis of resistance showed that CD8 T cells from PLP-immune mice transferred Qa-1-restricted inhibition of CD4 T cells responses to PLP responses, as judged by reduced expansion and IFN- responses of CD4 T cell populations.
The tempo of this inhibitory response may be contrasted with that of naturally occurring CD4+CD25+ regulatory T cells, which interrupt expansion of self-reactive T cell populations during the initial or innate stages of primary responses. Suppressive CD8 T cells arise late in the immune response, become apparent after restimulation by antigen or superantigen and act on antigen-activated CD4 T cells; CD4+CD25+ regulatory activity may inhibit unprimed T cells35,
36 or induce regulatory CD8 T cells37. The development of Qa-1-restricted suppressive activity more closely resembles that of CD8-dependent cytolytic activity rather than that of CD4+CD25+ regulatory activity. Both CD8 T cell−dependent activities require primary immunization to achieve either TCR-dependent CD8 CTL lysis of virally infected target cells or Qa-1-restricted functional elimination of a subpopulation of activated CD4 T cells23.
The mechanism of the Qa-1-restricted inhibitory interaction was not investigated here. CD8 T cells that recognize Qa-1 associated with the insulin peptide efficiently kill Qa-1-expressing lymphoblastoid target cells in the presence of intact soluble insulin; that is, after TAP-independent processing and presentation of insulin-derived peptide by Qa-1 on activated target cells25. The Qa-1-restricted inhibitory activity of CD8 T cells in vitro may reflect lysis of activated CD4 T cells that display Qa-1-peptide complexes, perhaps after TCR-dependent activation leading to expression of the relevant Qa-1 ligand on activated T and B lymphocytes. Although the precise Qa-1-binding peptides that target the CD8 responses described here have not been identified, they may represent a relatively restricted set that may be most efficiently expressed on activated T cells after TCR-ligand interactions of higher avidity23. The fate of individual TCRs after ligation with peptide-MHC complexes may depend on peptide affinity, the association and dissociation rate of TCR-ligand formation and the strength of TCR-dependent signals. Most internalized TCR complexes are degraded in lysozomes38, but robust TCR engagement may facilitate proteosome-dependent peptide degradation39 leading to recycling into a TAP-independent Qa-1 presentation pathway.
Although interactions between Qa-1 and CD94-NKG2A on CD8 T cells and NK or NKT cells can downregulate the cytotoxic activity of CD8 T cells, mice deficient in Qa-1 did not develop obvious defects in the development of CD4 and CD8 T cells. Analysis of Qa-1-restricted suppression by CD8 T cells in perforin-deficient hosts indicated a primary function for TCR-dependent rather than NKG2-dependent recognition. Expression of TCR was necessary for suppressive activity, as constraints placed on TCR repertoire expression resulted in loss of CD8-dependent suppressive activity; TCR is unlikely to contribute to CD8-mediated suppression, as less than 0.2% of NK1.1- CD8 T cells expressed this TCR (data not shown). However, further studies are needed to fully delineate the potential contribution of NKG2-containing receptors to suppressive activity expressed by CD8 T cells. Possibly the engagement of NKG2 receptors on suppressive CD8 T cells by Qa-1 may regulate the development and/or expression of suppressive activity by CD8 T cells that express a Qa-1-restricted TCR. Analysis of the CD8 suppressive activity from Qa-1 'knock-in' mice that express mutant Qa-1 molecules that differentially engage the TCR and NKG2 receptors may allow a molecular dissection of the contributions of these distinct Qa-1-dependent recognition events to suppressive CD8 activity.
Methods Mice. C57BL/6 RAG-2−perforin double-deficient mice were purchased from Taconic Laboratories. B6.Tlaa mice were provided by C. Aldrich (Indiana University, Evansville, Indiana); OT-1 Rag-1-deficient and OT-2 TCR-transgenic mice were provided by H. Ploegh (Harvard Medical School, Boston, Massachusetts) and were crossed with B6 Qa-1b-deficient mice. All experiments involving animals were done in compliance with federal laws and institutional guidelines and have been approved by the Dana Farber Cancer Institute Animal Care Use Committee.
Cell purification. CD4 and CD8 T cells were purified by negative selection. Single-cell suspensions of lymph node and spleen cells were incubated for 30 min with rat antibody to mouse CD8 (anti-CD8) or anti-CD4, in addition to anti-B220, anti-Mac-1, anti-Gr-1 and anti-NK1.1 (BD Pharmingen). After being washed, cells were incubated for 30 min with magnetic beads coated with sheep anti-rat (Dynal), and antibody-coated cells were removed by magnetic separation to isolate CD4 and CD8 T cells.
Ocular infection and HSK scores. Corneas of mice were scarified with a sterile 27-gauge needle followed by infection in the right eye with HSV-1 (KOS; 5 105 PFU) on day 0 and a second infection on day 1 at the same viral dose, as described40. The severity of clinical stromal keratitis was quantified based on the degree of corneal opacity: 1, 25 % of cornea; 2, 25−50%; 3, 50−75%; 4, 75−100%.
Adoptive transfer of CD8 T cells from SEA-immune mice. B6 and B6.Tlaa mice were injected intraperitoneally with 25 g SEA (Toxin Technologies) in PBS and were rechallenged intraperitoneally with 5 g SEA after 15 d. CD8 T cells were purified from lymph nodes and spleen 11 d later and were transferred into RAG-2−perforin-deficient mice (1 106 cells/mouse) along with CD4 T cells (1 106 cells/mouse) from nonimmune B6 or B6.Tlaa mice. Adoptive hosts were challenged with SEA (5 g given intraperitoneally) and were analyzed 5 d later.
Induction of EAE. EAE was induced by PLP peptide (amino acids 172−183; PVYIYFNTWTTC). Mice were immunized subcutaneously with 150 g of PLP peptide and 300 g of killed Mycobacterium tuberculosis (H37ev) in CFA distributed over three sites on the flanks. In addition, 400 ng of pertussis toxin (List Biological Laboratories) was injected intraperitoneally on days 0 and 2. In experiments analyzing EAE resistance, mice were given 10 g of PLP peptide in CFA without pertussis toxin 10−20 d before being exposed to PLP-CFA plus pertussis toxin to induce disease as described above. Disease scores were: 0, normal mouse with no sign of disease; 1, limp tail; 2, limp tail and partial hind limb weakness; 3, complete hind limb paralysis; 4, complete hind limb and partial front limb paralysis; and 5, moribund state or death.
Construction of lentivirus Qa-1b expression vector and infection of CD4 T cells. The pLenti6/V5 plasmid (Invitrogen) was digested with BamHI and SacII and ligated to the IRES2-EGFP fragment from pIRES2-EGFP (BD Bioscience Clontech) as a selection marker for the gene of interest. Full-length Qa-1 cDNA was amplified by RT-PCR from RNA isolated from C57/BL6 splenocytes with Superscript II reverse transcriptase and Pfu polymerase (Invitrogen). XhoI and BamHI restriction enzyme sites were introduced into the PCR primers for cloning into the same sites of pLenti-IRES-GFP and the sequence of Qa-1 cDNA was verified.
Lentiviral stocks were generated by transfection of the construct together with the packaging plasmids pLP1, pLP2 and pLP/VSVG (Invitrogen) with LipofectAmine 2000 (Invitrogen), according to the manufacturer's instructions. Virus was collected 72 h after transfection and titers were determined. CD4 T cells purified from B6 Qa-1-deficient mice were then infected with virus (at a multiplicity of infection of 5−10) plus 6 g/ml polybrene in 24-well plates (1 106 cells/well). Then the cells were spun at 2,000g for 45 min at 4 °C followed by incubation for 2 h at 37 °C and 5% CO2.
CD8 T cells from OT-2-immune mice. After activation of OT-2 splenic lymphocytes by ConA (5 g/ml) for 2 d to induce Qa-1b expression in culture, CD4 T cells were purified, irradiated (3,000 rads) and injected into B6 mice (2 106 cells given intraperitoneally). After 14 d, CD8 T cells purified from lymph node and spleen were adoptively transferred into RAG-2−perforin-deficient mice along with CD4 T cells purified from draining lymph nodes of OT-2 mice that had been immunized with 50 g of OVA peptide (amino acids 323−339; ISQAVHAAHAEINEAGR) in CFA 14 d earlier. Adoptive hosts were challenged with OVA peptide (50 g) in IFA and were analyzed 7 d later.
In vitro T cell restimulation. Draining lymph node cells from immunized mice (3 105 to 4 105 cells/well) were cultured with antigen and irradiated splenocytes from B6 mice (5 105 cell/well) in RPMI 1640 medium supplemented with 10% FCS and 50 M -mercaptoethanol. Cell proliferation was measured by [3H]thymidine incorporation (1 Ci/well) during the last 18 h of culture. Cytokine concentrations in supernatants were determined with an enzyme-linked immunosorbent assay (ELISA) kit (BD Pharmingen).
Received 10 December 2003; Accepted 20 February 2004; Published online: 18 April 2004.
REFERENCES
Cantor, H. T-cell receptor crossreactivity and autoimmune disease. Adv. Immunol.75, 209–233 (2000). | PubMed | ISI | ChemPort |
Bouneaud, C., Kourilsky, P. & Bousso, P. Impact of negative selection on the T cell repertoire reactive to a self-peptide: a large fraction of T cell clones escapes clonal deletion. Immunity13, 829–840 (2000). | Article | PubMed | ISI | ChemPort |
Goldrath, A.W. & Bevan, M.J. Selecting and maintaining a diverse T-cell repertoire. Nature402, 255–262 (1999). | Article | PubMed | ISI | ChemPort |
Slifka, M.K. et al. Preferential escape of subdominant CD8+ T cells during negative selection results in an altered antiviral T cell hierarchy. J. Immunol.170, 1231–1239 (2003). | PubMed | ISI | ChemPort |
Martin, D.A. et al. Defective CD95/APO-1/Fas signal complex formation in the human autoimmune lymphoproliferative syndrome, type Ia. Proc. Natl. Acad. Sci. USA96, 4552–4557 (1999). | Article | PubMed | ChemPort |
Kearney, E.R., Pape, K.A., Loh, D.Y. & Jenkins, M.K. Visualization of peptide-specific T cell immunity and peripheral tolerance induction in vivo. Immunity1, 327–339 (1994). | PubMed | ISI | ChemPort |
Anderton, S.M., Radu, C.G., Lowrey, P.A., Ward, E.S. & Wraith, D.C. Negative selection during the peripheral immune response to antigen. J. Exp. Med.193, 1–11 (2001). | Article | PubMed | ISI | ChemPort |
Panoutsakopoulou, V. et al. Analysis of the relationship between viral infection and autoimmune disease. Immunity15, 137–147 (2001). | Article | PubMed | ISI | ChemPort |
von Herrath, M.G. & Harrison, L.C. Antigen-induced regulatory T cells in autoimmunity. Nat. Rev. Immunol.3, 223–232 (2003). | Article | PubMed | ISI | ChemPort |
Gavin, M. & Rudensky, A. Control of immune homeostasis by naturally arising regulatory CD4+ T cells. Curr. Opin. Immunol.15, 690–696 (2003). | Article | PubMed | ISI | ChemPort |
Wood, K.J. & Sakaguchi, S. Regulatory T cells in transplantation tolerance. Nat. Rev. Immunol.3, 199–210 (2003). | Article | PubMed | ISI | ChemPort |
Noble, A., Zhao, Z.-S. & Cantor, H. Suppression of immune responses by CD8 cells: II. Qa-1 on activated B-cells stimulates CD8 cell suppression of Th2-dependent antibody responses. J. Immunol.160, 566–571 (1998). | PubMed | ISI | ChemPort |
Jiang, H. & Chess, L. The specific regulation of immune responses by CD8+ T cells restricted by the MHC class Ib molecule, Qa-1. Annu. Rev. Immunol.18, 185–216 (2000). | Article | PubMed | ISI | ChemPort |
Jiang, H., Zhang, S.-L. & Pernis, B. Role of CD8+ T cells in murine experimental allergic encephalomyelitis. Science256, 1213–1215 (1992). | PubMed | ISI | ChemPort |
Jiang, H. et al. T cell vaccination induces TCR V specific, Qa-1 restricted regulatory CD8+ T cells. Proc. Natl. Acad. Sci. USA95, 4533–4537 (1998). | Article | PubMed | ChemPort |
Aldrich, C.J. et al. T cell recognition of QA-1b antigens on cells lacking a functional Tap-2 transporter. J. Immunol.149, 3773–3777 (1992). | PubMed | ISI | ChemPort |
Lo, W.F., Ong, H., Metcalf, E.S. & Soloski, M.J. T cell responses to gram-negative intracellular bacterial pathogens: a role for CD8+ T cells in immunity to Salmonella infection and the involvement of MHC class Ib molecules. J. Immunol.162, 5398–5406 (1999). | PubMed | ISI | ChemPort |
Lo, W.F. et al. Molecular mimicry mediated by MHC class Ib molecules after infection with gram-negative pathogens. Nat. Med.6, 215–218 (2000). | Article | PubMed | ISI | ChemPort |
Sullivan, B.A., Kraj, P., Weber, D.A., Ignatowicz, L. & Jensen, P.E. Positive selection of a Qa-1-restricted T cell receptor with specificity for insulin. Immunity17, 95–105 (2002). | Article | PubMed | ISI | ChemPort |
Transy, C. et al. A low polymorphic mouse H-2 class I gene from the Tla complex is expressed in a broad variety of cell types. J. Exp. Med.166, 341–361 (1987). | PubMed | ISI |
Soloski, M.J., DeCloux, A., Aldrich, C.J. & Forman, J. Structural and functional characteristics of the class IB molecule, Qa-1. Immunol. Rev.147, 67–89 (1995). | PubMed | ISI | ChemPort |
2-microglobulin that binds to and presents peptides derived from self or foreign proteins
, preproinsulin, bacterial GroEL and HSP60 (refs. 
129) mice four generations to C57BL/6 mice for all of the following experimental analyses except those involving herpes stromal keratitis (HSK), for which we compared C57BL/6 
(IFN-
) and Qa-1 wild-type (
) mice infected in the right eye with HSV-1; disease was assessed on days 9 and 12 (refs. 
g/ml) plus irradiated wild-type splenocytes (3 
3+ CD4 T cell response
s.d. of three mice per group. Co-transfer of CD8 T cells from B6 or OT-1 RAG-2-deficient donors did not result in notable inhibition of the response of Qa-1-deficient OT-2 CD4 T cells: CD4 alone, 8.8 
is unlikely to contribute to CD8-mediated suppression, as less than 0.2% of NK1.1- CD8 T cells expressed this TCR (data not shown). However, further studies are needed to fully delineate the potential contribution of NKG2-containing receptors to suppressive activity expressed by CD8 T cells. Possibly the engagement of NKG2 receptors on suppressive CD8 T cells by Qa-1 may regulate the development and/or expression of suppressive activity by CD8 T cells that express a Qa-1-restricted TCR. Analysis of the CD8 suppressive activity from Qa-1 'knock-in' mice that express mutant Qa-1 molecules that differentially engage the TCR and NKG2 receptors may allow a molecular dissection of the contributions of these distinct Qa-1-dependent recognition events to suppressive CD8 activity.
25 % of cornea; 2, 25−50%; 3, 50−75%; 4, 75−100%.