Antigen-Specific CD8+ T Cells Mediate a Peptide-Induced Fatal Syndrome1
http://www.100md.com
免疫学杂志 2005年第11期
Abstract
Peptide immunotherapy both activates and suppresses the T cell response against known peptide Ags. Although pretreatment with VP2121–130 peptide inhibits the development of antiviral CTL specific for the immunodominant Db:VP2121–130 epitope expressed during acute Theiler’s murine encephalomyelitis virus infection, i.v. injection of this same peptide or MHC tetramers containing the peptide during an ongoing antiviral CTL response results in a peptide-induced fatal syndrome (PIFS) within 48 h. Susceptibility to PIFS is dependent on peptide-specific CD8+ T cells, varies among inbred strains of mice, and is not mediated by traditionally defined mechanisms of shock. Analyses using bone marrow chimeras and mutant mice demonstrate that susceptibility to PIFS is determined by the genotype of bone marrow-derived cells and requires the expression of perforin. Animals responding to peptide treatment with PIFS develop classical stress responses in the brain. These findings raise important considerations for the development of peptide therapies for active diseases to modify immune responses involving expanded populations of T cells. In summary, treatment with peptides or MHC-tetramers during a peptide-specific immune response can result in a fatal shock-like syndrome. Susceptibility to the syndrome is genetically determined, is mediated by CD8+ T cells, and requires expression of perforin. These findings raise concerns about the use of peptides and MHC tetramers in therapeutic schemes.
Introduction
The discovery that cell-mediated immunity focuses on peptide Ags led to the development of strategies for peptide-based immunotherapy. Peptides derived from allergens have been used to suppress CD4+ T cells important for promoting hypersensitivity reactions (1, 2, 3). Similarly, peptide treatments that inhibit T cells specific for graft-derived peptides presented by self MHC class I and class II molecules provides a basis for prolonging graft survival (4, 5). Whereas peptides have been used successfully to inhibit immune responses, attempts to activate T cells with peptide treatments after the disease process is established have not been as successful. Peptides administered in adjuvant stimulate a stronger CTL response toward some tumor Ags (6, 7, 8, 9, 10, 11, 12, 13). In other cases, peptide treatment induced unresponsiveness of tumor-specific CTL (14, 15, 16, 17). The physiological basis of these different outcomes is unknown. The preponderance of evidence indicates that it is more difficult to promote immunity toward known peptide:MHC epitopes than to inhibit peptide-specific responses.
We have sought to inhibit peptide-specific T cells in the Theiler’s murine encephalomyelitis virus (TMEV)3 model of multiple sclerosis. TMEV is a murine picornavirus that causes chronic inflammation, demyelination, and neurologic deficits in susceptible, but not resistant, strains of mice (18, 19). Resistant H-2b haplotype mice develop a robust CTL response in the CNS against a peptide (VP2122–130, FHAGSLLVFM), derived from a TMEV capsid protein by 7 days postinfection (20, 21). Using peptide:MHC tetramers, we demonstrated in H-2b haplotype mice that 35–70% of CD8+ T cells express TCRs specific for the Db:VP2121–130 epitope (22).
We have demonstrated that i.v. injection of VP2121–130 peptide before TMEV infection inhibits the appearance of Db:VP2121–130 epitope-specific CTL in the brain on day 7 and markedly protects the animals from ensuing paralytic disease (23, 24). Strategies that block peptide-specific CD8+ T cells before infection provide the immune system the opportunity to adapt by developing alternative responses to the Ag. Therefore, we sought to inhibit the response after infection and during active disease in an attempt to prevent the recruitment of alternative responses against the virus. Surprisingly, administration of VP2121–130 peptide during an ongoing antiviral CTL response had dramatically adverse effects. In C57BL/6 mice with an expanded Db:VP2121–130 epitope-specific CTL response, the administration of VP2121–130 peptide induced a fatal syndrome. Remarkably, even administration of MHC tetramers bearing the peptide in an attempt to image an ongoing immune response elicited the adverse response. This peptide-induced fatal syndrome (PIFS) is mediated by CD8+ T cells specific for the inciting peptide and is dependent on the expression of perforin. Furthermore, PIFS is not mediated by pathways previously defined in systemic shock. Susceptibility differs among mouse strains, is determined by the genotype of bone marrow-derived cells, and is inherited in a complex genetic fashion. The finding that peptide therapy or attempts to monitor Ag-specific responses can induce dramatically adverse effects provides important considerations for the development of schemes using peptides to suppress or monitor active immune responses that contain populations of expanded T cells.
Materials and Methods
Adoptive transfer of GFP splenocytes into C57BL/6 perforin–/– mice
C57BL/6 perforin–/– mice were irradiated with 400 rad. One day following irradiation, 108 GFP-expressing splenocytes were i.v. injected into the irradiated C57BL/6 perforin–/– mice. One day following GFP splenocyte transfer, mice were intracranially infected with TMEV. Eight days following TMEV infection, animals were administered 0.1 mg of VP2121–130 or E7 peptide. One day following peptide treatment, brains were excised, fixed in 4% paraformaldehyde, and cut with a vibratome. The sections were adhered to slides and the presence of GFP-expressing cells in both E7 and VP2121–130 peptide-treated groups was observed by fluorescent microscopy to confirm successful adoptive transfer.
Results
TMEV-infected mice develop anti-TMEV responses dominated by a single peptide epitope
CNS-infiltrating lymphocytes were isolated from the brains of several strains of H-2b haplotype mice 7 days postinfection with TMEV. Approximately 35–70% of CD8+ T cells isolated from brains of TMEV-infected C57BL/6, 129, TNF-–/–, and TNFRII–/– mice stained positive for both CD8 Ab and Db:VP2121–130 tetramer (Fig. 1, A-D), but not with a control tetramer bearing an irrelevant peptide derived from human papilloma virus (E). Also, as previously described, TMEV-infected IFN-R–/–, class II–/–, TNFRI–/–, and perforin–/– mice have a similar percentage of Db:VP2121–130 epitope-specific CD8+ T cells in their brains (22). CD4+ T cells did not express TCR specific for Db:VP2121–130 in these mice. As expected, few CNS-infiltrating lymphocytes were detected in the brains of uninfected C57BL/6 mice (22). These observations demonstrate that, at the peak of virus infection, many CD8+ T cells at the site of infection have TCRs specific for the VP2121–130 peptide presented in the context of Db.
Animals that received peptide i.v. 2 wk after virus infection were equally susceptible to PIFS (five of five animals succumbed). By 2 wk after infection, the number of T cells in the brains of infected mice is decreased by 90% relative to the numbers present 7 days after infection. This finding indicates that animals undergoing an active CD8+ response could be vulnerable over a prolonged period to systemic peptide treatments.
Administration of VP2121–130 peptide 8 days post-TMEV infection does not increase TMEV virus in IFN-R–/–, C57BL/6, and 129 mice undergoing VP2121–130 PIFS
The mechanism underlying the pathology of PIFS was not evident. We considered the possibility that a rapid and overwhelming rise in virus load might result from depletion of effective CTL by peptide treatment. To address this possibility, we prepared whole-brain RNA from IFN-R–/– mice 1 day after the VP2121–130 or E7 peptide treatment. Using probes complementary to TMEV RNA (VP2 capsid region) and to endogenous mouse GAPDH, we determined the ratio of virus to mouse RNA in individual animals. The ratio of virus to mouse GAPDH RNA present was not significantly different among the VP2121–130 (1.1 ± 0.04) and E7 (1.3 ± 0.8) peptide with a trend toward a lower virus load in VP2121–130 peptide-treated animals. In addition to the Northern blots, real-time RT-PCR was used to detect virus RNA load in C57BL/6 and 129 animals 1 day after VP2 and E7 peptide treatment. Consistent with IFN-R–/– mice, neither of these strains showed evidence of an increased virus load following VP2121–130 PIFS. These measurements ruled out the possibility that PIFS is the result of increased virus load.
PIFS is mediated by Ag-specific CD8+ T cells
We next considered the possibility that PIFS is mediated by activation of T cells by peptide treatment. To determine whether antiviral CD8+ T cells specific for the Db:VP2121–130 epitope were mediating PIFS observed in infected C57BL/6 and IFN-R–/– mice, the development of these T cells was inhibited by pretreating the mice with VP2121–130 peptide 1 day before TMEV infection. Pretreatment with VP2121–130 peptide results in a complete loss of Db:VP2121–130 epitope-specific CD8+ T cells in the brain at 7 days post-TMEV infection (Fig. 2B). Elimination of this specific set of CD8+ T cells prevented PIFS at day 8 post-TMEV infection (Table II). This result, observed in both C57BL/6 and IFN-R–/– mice, supports the hypothesis that VP2121–130 PIFS was dependant on the presence of a CD8+ response specific for the Db:VP2121–130 epitope in the CNS.
The finding that treatment of mice with peptide Ag during an ongoing class I-restricted response can lead to PIFS, led to the prediction that a transgenic mouse expressing an abundance of T cells bearing a single receptor might undergo a similar fate upon peptide treatment. Indeed, treatment of naive 2C transgenic mice i.v. with 0.1 mg of the peptide SIYRYYGL, a known ligand for the 2C receptor (30), induced a fatal syndrome clinically indistinguishable from PIFS in all eight animals tested. In contrast, none of the eight animals treated with the irrelevant Kb-binding peptide SIINFEKL developed any symptoms. Depletion of CD8+ T cells in 2C transgenic animals by systemic pretreatment with anti-CD8 Abs provided complete protection from PIFS (n = 5), providing further support for the hypothesis that Ag-specific CD8+ T cells mediate the syndrome.
PIFS is not class II dependent
Because rapid deterioration of the mice observed after VP2121–130 peptide treatment resembled symptoms observed in class II-restricted toxic shock, we addressed the possibility that CD4+ T cells were somehow involved in mediating PIFS. Class II–/–H-2 AB° mice cannot present the VP2121–130 peptide or any other peptide in the context of class II molecules and have greatly reduced numbers of CD4+ T cells, but still have high numbers of CNS-infiltrating CD8+ T cells specific for Db:VP2121–130 (22). Administration of the VP2121–130 peptide 8 days after TMEV infection resulted in PIFS in five of seven mice (Table I), demonstrating that the mechanisms mediating VP2121–130 PIFS are not dependent on class II presentation of peptide.
It is possible that a specific gene could confer resistance or susceptibility to VP2121–130 PIFS. To address this possibility, (129 x B6)F1 mice were backcrossed with B6 mice. Of the 60 (129 x B6) x B6 mice generated through this backcross, 28 succumbed to PIFS when administered the VP2 peptide 8 days post-TMEV infection (Table IV). The observed 1:1 ratio of the backcross animals succumbing to VP2121–130 PIFS is in accordance with the hypothesis that a single gene confers resistance or susceptibility to VP2121–130 PIFS. However, microsatellite analysis performed on animals generated in this backcross revealed that susceptibility to VP2121–130 PIFS was dictated in a complex fashion and identified no single gene that can account for the resistance of 129 mice. However, C57BL/6 genes located near D2Mit312 (p = 0.0009) and D16Mit34 (p = 0.0039) on chromosomes 2 and 16 were both found to render mice more resistant to VP2121–130 PIFS. The importance of these loci was confirmed by two independent analyses. This was an unexpected result, considering that the C57BL/6 strain is susceptible to VP2121–130 PIFS. An implication of these findings is that genes are interacting to determine susceptibility to PIFS and that susceptibility to PIFS is under complex genetic control.
It should be noted that some chromosomal segments could not be analyzed using this procedure because of the highly similar microsatellite sequences present in the analyzed mouse lines. We attempted to specifically look at the importance of genes mapping to the region of chromosome 10 that includes the perforin locus, but were unsuccessful because no polymorphisms in this region were discernable. Even a search for single-nucleotide differences within the amplified segments failed to identify differences in this region of chromosome 10.
The question of whether genetic differences defining susceptibility to PIFS influence the properties of the immune system or influence the differential susceptibility of vital tissues to immune mediators was addressed using bone marrow chimera experiments. Lethally irradiated, resistant 129 mice were reconstituted with bone marrow from susceptible C57BL/6-ly5.1 mice. Six weeks after chimerization, the animals were challenged with TMEV intracranially. On day 7 postinfection, the animals were treated with 100 μg of VP2121–130 peptide i.v. and scored for PIFS over the next 2 days. In contrast to the completely resistant phenotype of nonchimeric 129 mice to PIFS, seven of eight C57BL/6-ly5.1 129 chimeric animals succumbed to PIFS (p < 0.001). Analysis of the brain-infiltrating lymphocytes by flow cytometry revealed that >95% of the T cells in the brains of the TMEV-infected chimeric mice were of donor (susceptible phenotype) origin. This finding is consistent with the hypothesis that the genetic differences distinguishing resistance and susceptibility in C57BL/6 and 129 mice is a property of bone marrow-derived cells and likely is attributable to differences in the immune response. The reciprocal experiment was also performed in which resistant 129 bone marrow was used to reconstitute C57BL/6-ly5.1 animals. However, in this case, chimerism was incomplete with only 40–60% of the brain-infiltrating lymphocytes being of donor origin. This group of chimeric mice was somewhat more resistant (four of eight animals succumbed to PIFS) in comparison to the uniformly susceptible C57BL/6-ly.1 and standard C57BL/6 mice. The incomplete protection of the partially chimeric animals can be attributed to the substantial numbers of host T cells present in the TMEV-infected brains. Again, these results, although not definitive, are consistent with the view that susceptibility to PIFS is a determinant associated with traits expressed by C57BL/6 T cells.
The finding that 129 mice are resistant to PIFS permitted the analysis of the fate of VP2121–130-specific T cells following treatment of mice with peptide at the height of the inflammatory response in the brain. We noted an equivalent reduction (by 30%) in the number of T cells 1 day after peptide treatment in both 129 and C57BL/6 animals. Whereas C57BL/6 mice succumb to PIFS by day 2, 129 animals survive indefinitely. By 4 days after peptide treatment, the number of VP2121–130-positive T cells in the 129 mice had returned to normal. We also examined the possibility that the specificity of responding CD8+ T cells infiltrating the brains of infected B6 and 129 mice might be qualitatively different. We first compared the avidity of T cells for the viral peptide VP2121–131 presented in the context of the Db class I molecule. Db-tetramers bearing the VP2 peptide were bound to brain-infiltrating T cells isolated from 129 and C57BL/6-virus-infected animals. Disassociation of labeled tetramers was assessed over time following blocking of their ability to reassociate with unlabeled tetramer. No differences in disassociation rates were observed in our analysis of 129 and C57BL/6 virus-specific T cells (data not shown). We next measured the direct cytolytic activity of the brain-infiltrating lymphocytes isolated from the brains of 8-day TMEV-infected B6 and 129 mice 1 day after VP2 peptide treatment. CTL from both strains showed equivalent ability to lyse C57SV targets expressing Db and the VP2 protein in a standard 51Cr release assay (data not shown). Therefore, we found no evidence of increased apoptotic activity, differences in receptor repertoires, or changes in cytolytic activity that might account for the relative susceptibility of C57BL/6 mice in comparison to PIFS-resistant 129 mice.
Administration of VP2121–130 at 8 days post-TMEV infection results in blood vessel permeability and tissue abnormality in the brain
Autopsies were performed on TMEV-infected C57BL/6 mice 1 day after injection of E7 or VP2121–130 peptide. Mice given VP2121–130 but not E7 peptide were moribund when tissue was harvested. Using H&E staining of frozen sections, we observed no changes in the liver, kidney, spleen, lung, lymph node, thymus, or skeletal muscle of mice treated with VP2121–130 peptide when compared with organs of E7-treated controls not undergoing PIFS (data not shown). We concluded that the peptide-induced changes did not result in direct systemic damage, a finding consistent with our previous studies showing that peptide-specific T cells are largely colocalized in the CNS of infected animals (22).
Because PIFS resembled septic shock, we explored whether blood vessel permeability could be associated with the syndrome. C57BL/6 mice were injected i.v. with VP2121–130 8 days after TMEV infection to initiate PIFS. The following day, VP2121–130- or E7-treated mice were injected i.v. with HRP. One hour later, brain tissue was harvested for histological analysis. In comparison to brain tissue from mice treated with irrelevant E7 peptide (Fig. 3A), brains of mice succumbing to PIFS revealed extensive diffusion of HRP throughout the brain (B). We conclude that an extensive loss of vascular integrity occurs in the brains of mice with PIFS. No vascular leakage of HRP was observed in the heart, kidneys, lungs, liver, and peritoneum of these mice (data not shown). Furthermore, H&E staining of brain sections revealed extravasation of RBC in animals undergoing PIFS, but not in E7-treated control animals. In particular, extravascular erythrocytes were observed in the hypothalamus, cortex, hippocampus, striatum, and corpus callosum from moribund mice (Fig. 3, C and D). These experiments demonstrate that i.v. VP2121–130 peptide treatment induced blood vessel permeability in the brain.
In a separate experiment, we injected i.v. superparamagnetic Db:VP2121–130 tetramer into 7-day TMEV-infected animals to determine whether brain-infiltrating Db:VP2121–130 epitope-specific CD8+ T cells could be visualized using MRI. Injection of i.v. superparamagnetic Db:VP2121–130 tetramer induced PIFS in these animals. Moreover, MRI revealed extensive abnormalities in the CNS of tetramer-treated animals as demonstrated by T2-weighted images. As shown in Fig. 3F, the corpus callosum, hippocampus, cortex, thalamus, and striatum show evidence of tissue abnormality (normal tissue is dark in this analysis, whereas inflamed areas of tissue are light or washed out) 1 day following the administration of superparamagnetic Db:VP2121–130 tetramer. No such abnormality was observed in animals administered an irrelevant superparamagnetic Db:E7 tetramer (Fig. 3E). This demonstrated that, in addition to vessel permeability, there is extensive damage to CNS tissue following the induction of VP2121–130 PIFS with superparamagnetic Db:VP2121–130 tetramer, suggesting that such technology to image specific lymphocytes could be potentially hazardous to use clinically.
Our findings of tissue damage in the brain by MRI and gross histological analysis is further supported by the activation of glia following systemic treatment with antigenic peptide. Up-regulated expression of glial fibrillary acidic protein, a pattern associated with the response of astrocytes to immunologic and traumatic injury, indicates that systemic peptide treatment leads to a widespread stress response in brain cells (Fig. 4) (34, 35). Tissue damage in the brain is normally associated with the influx of macrophages and the activation of resident microglial cells. The widespread appearance of cells bearing the macrophage/microglia marker F4/80 in the brains of C57BL/6 mice with PIFS is a marker for the onset of these inflammatory changes associated with damage in the brain. In contrast, TMEV-infected 129 mice treated systemically with peptide showed very little and localized evidence of macrophage infiltration or microglial activation (compare top and bottom right panels) and essentially no stress response by astrocytes (compare top and bottom left panels). This is further evidence that there are strong strain-dependent injury responses in the brain to systemic peptide treatment of mice undergoing an active T cell response.
Because perforin–/– mice were generated on the 129 genetic background, it remained possible that 129 genes carried from the original perforin–/– founder mouse could be responsible for resistance to VP2121–130 PIFS. To rule out background genes playing a role, we adoptively transferred perforin-competent GFP-expressing splenocytes into irradiated perforin–/– mice 1 day before TMEV infection. Following transfer, >50% of brain-infiltrating CD8+ T cells during acute TMEV infection expressed GFP and hence also expressed perforin (data not shown). Perforin–/– mice reconstituted with GFP and perforin-expressing splenocytes readily succumbed to VP2121–130 PIFS, with seven of eight animals becoming symptomatic or moribund within 24 h with systematic peptide treatment (Table V). Our analysis of the animals revealed that both VP2121–130 and E7 peptide-treated animals had GFP-expressing cells in the cortex, striatum, hypothalamus, corpus callosum, and hippocampus (Fig. 5, E and F; data not shown). This experiment demonstrates that VP2121–130 PIFS is dependent on the expression of perforin by CNS-infiltrating immune cells and rules out the possibility that the 129 genes conferring resistance to this syndrome influence perforin release by the peptide-specific T cells.
Discussion
These experiments demonstrate that i.v. administration of class I-restricted peptides or class I tetramers that contain these peptides can be fatal to individuals harboring expanded, peptide-specific CD8+ populations. There is one well-characterized mechanism by which a response can lead to rapid death. Pathogen-derived superantigens simultaneously bind both the receptor (TCR) V chain and class II molecules (36). This cross-linking of TCR with class II leads to the polyclonal activation of CD4+ Th cells followed by the systemic increase of cytokines. Shock results from a dramatic rise in the cytokines TNF-, TNF-, IL-1, IL-6, IL-8, and IL-10. These cytokines initiate a cascade of fever, increased cardiac output, increased blood clotting, organ failure, and death (31). Our analysis rules out these classical shock pathways. In this study, we address a second T cell-mediated syndrome that is initiated by peptide-specific CD8+ T cells. Although CD4+ T cell-mediated shock is becoming better understood, the possibility of CD8+ T cells mediating similar syndromes has gone almost completely unnoticed. One study describing the possibility of CD8+ T cell-mediated shock used Abs that initiate polyclonal activation and their secretion of IFN- and TNF-. These cytokines were protective in the host response to Trypanosoma cruzi. However, when these T. cruzi-infected mice were administered anti-CD3 Ab, these mice underwent septic shock. Pretreatment with anti-CD8 Abs before this anti-CD3 treatment prevented septic shock (37). These observations suggested that, in this disease model, septic shock might be mediated by CD8+ T cells.
The importance of CD8+ T cells in inducing shock was suggested in other studies involving LCMV and tumor peptide vaccination (33, 38, 39). In the LCMV model, administration of peptide activated peptide-specific memory cells, inducing damage in the spleen that was generally immunosuppressive (38). Shock induced with LCMV was mediated by lymphotoxin as recombinant soluble chimeric lymphotoxin -receptor/Ab (LT-R) blocked virus-induced pathology (33). In the tumor peptide vaccination model, CD8+ T cell-mediated shock was mediated by TNF because anti-TNF Abs neutralized this fatal condition (39). In this study, we demonstrate that a fatal syndrome can be induced by the treatment of virus-infected mice with soluble peptide. Mice succumbing to PIFS display a dramatic breakdown of vascular integrity in the brain. In our system, TNF-, TNFRI, and TNFRII are not required for the development of the PIFS. Furthermore, induction of PIFS also is not influenced by treating animals with soluble LT-R fusion protein, indicating that none of the previously defined cytokine pathways appears to play a dominant role in this syndrome. Another possibility is that CD8+ T cells secrete cytokines that directly or indirectly lead to vessel permeability. Vessel permeability caused by cytokines has been reported by those investigating vascular leak syndrome (VLS), a condition where vascular epithelial cells become damaged and permeable to vascular fluids (40). In particular, the administration of rIL-2, a commonly emitted CD8+ cytokine, can initiate murine VLS (40, 41, 42). Although VLS is a trait of VP2121–130 PIFS, permeability of blood vessels in the brain alone does not lead to death. Perforin-deficient mice used in this study have extensive vessel permeability in the CNS comparable to C57BL/6 controls, but remain asymptomatic. Therefore, we hypothesize that perforin destabilizes cells in the CNS involved with vital functions. This is consistent with previous observations that perforin contributes to pathogenesis (28, 29, 43, 44, 45). Therefore, we may be triggering a CD8+ T cell-mediated process that occurs naturally following infection in the CNS.
The observation that CD8+ T cells can mediate a PIFS extends beyond the Theiler’s virus system. We have investigated whether peptide could induce shock in the 2C transgenic mouse. 2C transgenic mice express receptors that are reactive with the SIYR peptide presented in the context of Kb class I molecules on >90% of peripheral T cells. The effect of injecting SIYR peptide i.v. into healthy 2C TCR transgenic mice was similar to that observed upon VP2121–130 peptide treatment of 8-day TMEV-infected C57BL/6, class II–/–, and IFN-R–/– mice. PIFS was not reproduced with an irrelevant Kb-binding OVA peptide, demonstrating the peptide specificity of the syndrome. As little as 0.001 mg of SIYR peptide administered three times i.v. could induce PIFS in 2C transgenic mice. Similar to the VP2121–130 peptide-induced syndrome, SIYR PIFS was mediated by an expanded population of epitope-specific CD8+ T cells.
Peptide-specific CD8+ T cells are necessary, but not sufficient, to render animals susceptible to PIFS. Besides expression of perforin, PIFS is also influenced by the genetic makeup of the host. More specifically, we present evidence that resistance to PIFS is conferred through a complex interaction of multiple genes. The influence of mouse genotype can, therefore, explain how 129 mice containing populations of Db:VP2121–130 epitope-specific CD8+ T cells equivalent to C57BL/6 mice can remain resistant to VP2121–130 PIFS. More work is needed to determine whether mouse genotype influences perforin function. Because it is likely that peptide-mediated suppression will continue to be a strategy to down-regulate subsets of Ag-specific T cells, we find these observations to be highly relevant to those interested in the development of peptide therapies.
Acknowledgments
We thank John Altman of Emory University (Atlanta, GA) and Mark Davis of Stanford University (Stanford, CA) for contributing the Db and human 2m expression plasmids. We also thank Lieping Chen of Mayo for soluble LT-R. We also thank Mike Bell, Charles Howe, Kevin Pavelko, Loc Nguyen, and Yanice Mendez-Fernandez for technical assistance in the completion of this work.
Footnotes
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.
1 This work was funded by National Institutes of Health Grants N524180 and N532129.
2 Address correspondence and reprint requests to Dr. Larry R. Pease, Department of Immunology, Mayo Clinic College of Medicine, 200 First Street SW, Rochester, MN 55905. E-mail address: pease.larry{at}mayo.edu
3 Abbreviations used in this paper: TMEV, Theiler’s murine encephalomyelitis virus; PIFS, peptide-induced fatal syndrome; MRI, magnetic resonance imaging; LCMV, lymphocytic choriomeningitis virus; LT, lymphotoxin ; VLS, vascular leak syndrome.
Received for publication November 17, 2004. Accepted for publication February 18, 2005.
References
Hoyne, G. F., R. E. O’Hehir, D. C. Wraith, W. R. Thomas, J. R. Lamb. 1993. Inhibition of and antibody responses to house dust mite allergen by inhalation of the dominant epitope in naive and sensitized mice. J. Exp. Med. 178: 1783-1788.
Rolland, J., R. O’Hehih. 1998. Immunotherapy of allergy: anergy, deletion, and immune deviation. Curr. Opin. Immunol. 10: 640-645.
Muller, U., C. A. Akdis, M. Fricker, M. Akdis, T. Blesken, F. Bettens, K. Blaser. 1998. Successful immunotherapy with T-cell epitope peptides of bee venom phospholipase A2 induces specific T-cell anergy in patients allergic to bee venom. J. Allergy Clin. Immunol. 101: 747-754.
Magee, C. C., M. H. Sayegh. 1998. Peptide mediated immunosuppression: new developments. Transplant. Proc. 30: 2131-2135.
Magee, C. C., M. H. Sayegh. 1997. Peptide-mediated immunosuppression. Curr. Opin. Immunol. 9: 669-675.
Minev, B. R., F. L. Chavez, M. S. Mitchell. 1998. New trends in development of cancer vaccines. In Vivo 12: 629-638.(Aaron J. Johnson*,, Yanic)
Peptide immunotherapy both activates and suppresses the T cell response against known peptide Ags. Although pretreatment with VP2121–130 peptide inhibits the development of antiviral CTL specific for the immunodominant Db:VP2121–130 epitope expressed during acute Theiler’s murine encephalomyelitis virus infection, i.v. injection of this same peptide or MHC tetramers containing the peptide during an ongoing antiviral CTL response results in a peptide-induced fatal syndrome (PIFS) within 48 h. Susceptibility to PIFS is dependent on peptide-specific CD8+ T cells, varies among inbred strains of mice, and is not mediated by traditionally defined mechanisms of shock. Analyses using bone marrow chimeras and mutant mice demonstrate that susceptibility to PIFS is determined by the genotype of bone marrow-derived cells and requires the expression of perforin. Animals responding to peptide treatment with PIFS develop classical stress responses in the brain. These findings raise important considerations for the development of peptide therapies for active diseases to modify immune responses involving expanded populations of T cells. In summary, treatment with peptides or MHC-tetramers during a peptide-specific immune response can result in a fatal shock-like syndrome. Susceptibility to the syndrome is genetically determined, is mediated by CD8+ T cells, and requires expression of perforin. These findings raise concerns about the use of peptides and MHC tetramers in therapeutic schemes.
Introduction
The discovery that cell-mediated immunity focuses on peptide Ags led to the development of strategies for peptide-based immunotherapy. Peptides derived from allergens have been used to suppress CD4+ T cells important for promoting hypersensitivity reactions (1, 2, 3). Similarly, peptide treatments that inhibit T cells specific for graft-derived peptides presented by self MHC class I and class II molecules provides a basis for prolonging graft survival (4, 5). Whereas peptides have been used successfully to inhibit immune responses, attempts to activate T cells with peptide treatments after the disease process is established have not been as successful. Peptides administered in adjuvant stimulate a stronger CTL response toward some tumor Ags (6, 7, 8, 9, 10, 11, 12, 13). In other cases, peptide treatment induced unresponsiveness of tumor-specific CTL (14, 15, 16, 17). The physiological basis of these different outcomes is unknown. The preponderance of evidence indicates that it is more difficult to promote immunity toward known peptide:MHC epitopes than to inhibit peptide-specific responses.
We have sought to inhibit peptide-specific T cells in the Theiler’s murine encephalomyelitis virus (TMEV)3 model of multiple sclerosis. TMEV is a murine picornavirus that causes chronic inflammation, demyelination, and neurologic deficits in susceptible, but not resistant, strains of mice (18, 19). Resistant H-2b haplotype mice develop a robust CTL response in the CNS against a peptide (VP2122–130, FHAGSLLVFM), derived from a TMEV capsid protein by 7 days postinfection (20, 21). Using peptide:MHC tetramers, we demonstrated in H-2b haplotype mice that 35–70% of CD8+ T cells express TCRs specific for the Db:VP2121–130 epitope (22).
We have demonstrated that i.v. injection of VP2121–130 peptide before TMEV infection inhibits the appearance of Db:VP2121–130 epitope-specific CTL in the brain on day 7 and markedly protects the animals from ensuing paralytic disease (23, 24). Strategies that block peptide-specific CD8+ T cells before infection provide the immune system the opportunity to adapt by developing alternative responses to the Ag. Therefore, we sought to inhibit the response after infection and during active disease in an attempt to prevent the recruitment of alternative responses against the virus. Surprisingly, administration of VP2121–130 peptide during an ongoing antiviral CTL response had dramatically adverse effects. In C57BL/6 mice with an expanded Db:VP2121–130 epitope-specific CTL response, the administration of VP2121–130 peptide induced a fatal syndrome. Remarkably, even administration of MHC tetramers bearing the peptide in an attempt to image an ongoing immune response elicited the adverse response. This peptide-induced fatal syndrome (PIFS) is mediated by CD8+ T cells specific for the inciting peptide and is dependent on the expression of perforin. Furthermore, PIFS is not mediated by pathways previously defined in systemic shock. Susceptibility differs among mouse strains, is determined by the genotype of bone marrow-derived cells, and is inherited in a complex genetic fashion. The finding that peptide therapy or attempts to monitor Ag-specific responses can induce dramatically adverse effects provides important considerations for the development of schemes using peptides to suppress or monitor active immune responses that contain populations of expanded T cells.
Materials and Methods
Adoptive transfer of GFP splenocytes into C57BL/6 perforin–/– mice
C57BL/6 perforin–/– mice were irradiated with 400 rad. One day following irradiation, 108 GFP-expressing splenocytes were i.v. injected into the irradiated C57BL/6 perforin–/– mice. One day following GFP splenocyte transfer, mice were intracranially infected with TMEV. Eight days following TMEV infection, animals were administered 0.1 mg of VP2121–130 or E7 peptide. One day following peptide treatment, brains were excised, fixed in 4% paraformaldehyde, and cut with a vibratome. The sections were adhered to slides and the presence of GFP-expressing cells in both E7 and VP2121–130 peptide-treated groups was observed by fluorescent microscopy to confirm successful adoptive transfer.
Results
TMEV-infected mice develop anti-TMEV responses dominated by a single peptide epitope
CNS-infiltrating lymphocytes were isolated from the brains of several strains of H-2b haplotype mice 7 days postinfection with TMEV. Approximately 35–70% of CD8+ T cells isolated from brains of TMEV-infected C57BL/6, 129, TNF-–/–, and TNFRII–/– mice stained positive for both CD8 Ab and Db:VP2121–130 tetramer (Fig. 1, A-D), but not with a control tetramer bearing an irrelevant peptide derived from human papilloma virus (E). Also, as previously described, TMEV-infected IFN-R–/–, class II–/–, TNFRI–/–, and perforin–/– mice have a similar percentage of Db:VP2121–130 epitope-specific CD8+ T cells in their brains (22). CD4+ T cells did not express TCR specific for Db:VP2121–130 in these mice. As expected, few CNS-infiltrating lymphocytes were detected in the brains of uninfected C57BL/6 mice (22). These observations demonstrate that, at the peak of virus infection, many CD8+ T cells at the site of infection have TCRs specific for the VP2121–130 peptide presented in the context of Db.
Animals that received peptide i.v. 2 wk after virus infection were equally susceptible to PIFS (five of five animals succumbed). By 2 wk after infection, the number of T cells in the brains of infected mice is decreased by 90% relative to the numbers present 7 days after infection. This finding indicates that animals undergoing an active CD8+ response could be vulnerable over a prolonged period to systemic peptide treatments.
Administration of VP2121–130 peptide 8 days post-TMEV infection does not increase TMEV virus in IFN-R–/–, C57BL/6, and 129 mice undergoing VP2121–130 PIFS
The mechanism underlying the pathology of PIFS was not evident. We considered the possibility that a rapid and overwhelming rise in virus load might result from depletion of effective CTL by peptide treatment. To address this possibility, we prepared whole-brain RNA from IFN-R–/– mice 1 day after the VP2121–130 or E7 peptide treatment. Using probes complementary to TMEV RNA (VP2 capsid region) and to endogenous mouse GAPDH, we determined the ratio of virus to mouse RNA in individual animals. The ratio of virus to mouse GAPDH RNA present was not significantly different among the VP2121–130 (1.1 ± 0.04) and E7 (1.3 ± 0.8) peptide with a trend toward a lower virus load in VP2121–130 peptide-treated animals. In addition to the Northern blots, real-time RT-PCR was used to detect virus RNA load in C57BL/6 and 129 animals 1 day after VP2 and E7 peptide treatment. Consistent with IFN-R–/– mice, neither of these strains showed evidence of an increased virus load following VP2121–130 PIFS. These measurements ruled out the possibility that PIFS is the result of increased virus load.
PIFS is mediated by Ag-specific CD8+ T cells
We next considered the possibility that PIFS is mediated by activation of T cells by peptide treatment. To determine whether antiviral CD8+ T cells specific for the Db:VP2121–130 epitope were mediating PIFS observed in infected C57BL/6 and IFN-R–/– mice, the development of these T cells was inhibited by pretreating the mice with VP2121–130 peptide 1 day before TMEV infection. Pretreatment with VP2121–130 peptide results in a complete loss of Db:VP2121–130 epitope-specific CD8+ T cells in the brain at 7 days post-TMEV infection (Fig. 2B). Elimination of this specific set of CD8+ T cells prevented PIFS at day 8 post-TMEV infection (Table II). This result, observed in both C57BL/6 and IFN-R–/– mice, supports the hypothesis that VP2121–130 PIFS was dependant on the presence of a CD8+ response specific for the Db:VP2121–130 epitope in the CNS.
The finding that treatment of mice with peptide Ag during an ongoing class I-restricted response can lead to PIFS, led to the prediction that a transgenic mouse expressing an abundance of T cells bearing a single receptor might undergo a similar fate upon peptide treatment. Indeed, treatment of naive 2C transgenic mice i.v. with 0.1 mg of the peptide SIYRYYGL, a known ligand for the 2C receptor (30), induced a fatal syndrome clinically indistinguishable from PIFS in all eight animals tested. In contrast, none of the eight animals treated with the irrelevant Kb-binding peptide SIINFEKL developed any symptoms. Depletion of CD8+ T cells in 2C transgenic animals by systemic pretreatment with anti-CD8 Abs provided complete protection from PIFS (n = 5), providing further support for the hypothesis that Ag-specific CD8+ T cells mediate the syndrome.
PIFS is not class II dependent
Because rapid deterioration of the mice observed after VP2121–130 peptide treatment resembled symptoms observed in class II-restricted toxic shock, we addressed the possibility that CD4+ T cells were somehow involved in mediating PIFS. Class II–/–H-2 AB° mice cannot present the VP2121–130 peptide or any other peptide in the context of class II molecules and have greatly reduced numbers of CD4+ T cells, but still have high numbers of CNS-infiltrating CD8+ T cells specific for Db:VP2121–130 (22). Administration of the VP2121–130 peptide 8 days after TMEV infection resulted in PIFS in five of seven mice (Table I), demonstrating that the mechanisms mediating VP2121–130 PIFS are not dependent on class II presentation of peptide.
It is possible that a specific gene could confer resistance or susceptibility to VP2121–130 PIFS. To address this possibility, (129 x B6)F1 mice were backcrossed with B6 mice. Of the 60 (129 x B6) x B6 mice generated through this backcross, 28 succumbed to PIFS when administered the VP2 peptide 8 days post-TMEV infection (Table IV). The observed 1:1 ratio of the backcross animals succumbing to VP2121–130 PIFS is in accordance with the hypothesis that a single gene confers resistance or susceptibility to VP2121–130 PIFS. However, microsatellite analysis performed on animals generated in this backcross revealed that susceptibility to VP2121–130 PIFS was dictated in a complex fashion and identified no single gene that can account for the resistance of 129 mice. However, C57BL/6 genes located near D2Mit312 (p = 0.0009) and D16Mit34 (p = 0.0039) on chromosomes 2 and 16 were both found to render mice more resistant to VP2121–130 PIFS. The importance of these loci was confirmed by two independent analyses. This was an unexpected result, considering that the C57BL/6 strain is susceptible to VP2121–130 PIFS. An implication of these findings is that genes are interacting to determine susceptibility to PIFS and that susceptibility to PIFS is under complex genetic control.
It should be noted that some chromosomal segments could not be analyzed using this procedure because of the highly similar microsatellite sequences present in the analyzed mouse lines. We attempted to specifically look at the importance of genes mapping to the region of chromosome 10 that includes the perforin locus, but were unsuccessful because no polymorphisms in this region were discernable. Even a search for single-nucleotide differences within the amplified segments failed to identify differences in this region of chromosome 10.
The question of whether genetic differences defining susceptibility to PIFS influence the properties of the immune system or influence the differential susceptibility of vital tissues to immune mediators was addressed using bone marrow chimera experiments. Lethally irradiated, resistant 129 mice were reconstituted with bone marrow from susceptible C57BL/6-ly5.1 mice. Six weeks after chimerization, the animals were challenged with TMEV intracranially. On day 7 postinfection, the animals were treated with 100 μg of VP2121–130 peptide i.v. and scored for PIFS over the next 2 days. In contrast to the completely resistant phenotype of nonchimeric 129 mice to PIFS, seven of eight C57BL/6-ly5.1 129 chimeric animals succumbed to PIFS (p < 0.001). Analysis of the brain-infiltrating lymphocytes by flow cytometry revealed that >95% of the T cells in the brains of the TMEV-infected chimeric mice were of donor (susceptible phenotype) origin. This finding is consistent with the hypothesis that the genetic differences distinguishing resistance and susceptibility in C57BL/6 and 129 mice is a property of bone marrow-derived cells and likely is attributable to differences in the immune response. The reciprocal experiment was also performed in which resistant 129 bone marrow was used to reconstitute C57BL/6-ly5.1 animals. However, in this case, chimerism was incomplete with only 40–60% of the brain-infiltrating lymphocytes being of donor origin. This group of chimeric mice was somewhat more resistant (four of eight animals succumbed to PIFS) in comparison to the uniformly susceptible C57BL/6-ly.1 and standard C57BL/6 mice. The incomplete protection of the partially chimeric animals can be attributed to the substantial numbers of host T cells present in the TMEV-infected brains. Again, these results, although not definitive, are consistent with the view that susceptibility to PIFS is a determinant associated with traits expressed by C57BL/6 T cells.
The finding that 129 mice are resistant to PIFS permitted the analysis of the fate of VP2121–130-specific T cells following treatment of mice with peptide at the height of the inflammatory response in the brain. We noted an equivalent reduction (by 30%) in the number of T cells 1 day after peptide treatment in both 129 and C57BL/6 animals. Whereas C57BL/6 mice succumb to PIFS by day 2, 129 animals survive indefinitely. By 4 days after peptide treatment, the number of VP2121–130-positive T cells in the 129 mice had returned to normal. We also examined the possibility that the specificity of responding CD8+ T cells infiltrating the brains of infected B6 and 129 mice might be qualitatively different. We first compared the avidity of T cells for the viral peptide VP2121–131 presented in the context of the Db class I molecule. Db-tetramers bearing the VP2 peptide were bound to brain-infiltrating T cells isolated from 129 and C57BL/6-virus-infected animals. Disassociation of labeled tetramers was assessed over time following blocking of their ability to reassociate with unlabeled tetramer. No differences in disassociation rates were observed in our analysis of 129 and C57BL/6 virus-specific T cells (data not shown). We next measured the direct cytolytic activity of the brain-infiltrating lymphocytes isolated from the brains of 8-day TMEV-infected B6 and 129 mice 1 day after VP2 peptide treatment. CTL from both strains showed equivalent ability to lyse C57SV targets expressing Db and the VP2 protein in a standard 51Cr release assay (data not shown). Therefore, we found no evidence of increased apoptotic activity, differences in receptor repertoires, or changes in cytolytic activity that might account for the relative susceptibility of C57BL/6 mice in comparison to PIFS-resistant 129 mice.
Administration of VP2121–130 at 8 days post-TMEV infection results in blood vessel permeability and tissue abnormality in the brain
Autopsies were performed on TMEV-infected C57BL/6 mice 1 day after injection of E7 or VP2121–130 peptide. Mice given VP2121–130 but not E7 peptide were moribund when tissue was harvested. Using H&E staining of frozen sections, we observed no changes in the liver, kidney, spleen, lung, lymph node, thymus, or skeletal muscle of mice treated with VP2121–130 peptide when compared with organs of E7-treated controls not undergoing PIFS (data not shown). We concluded that the peptide-induced changes did not result in direct systemic damage, a finding consistent with our previous studies showing that peptide-specific T cells are largely colocalized in the CNS of infected animals (22).
Because PIFS resembled septic shock, we explored whether blood vessel permeability could be associated with the syndrome. C57BL/6 mice were injected i.v. with VP2121–130 8 days after TMEV infection to initiate PIFS. The following day, VP2121–130- or E7-treated mice were injected i.v. with HRP. One hour later, brain tissue was harvested for histological analysis. In comparison to brain tissue from mice treated with irrelevant E7 peptide (Fig. 3A), brains of mice succumbing to PIFS revealed extensive diffusion of HRP throughout the brain (B). We conclude that an extensive loss of vascular integrity occurs in the brains of mice with PIFS. No vascular leakage of HRP was observed in the heart, kidneys, lungs, liver, and peritoneum of these mice (data not shown). Furthermore, H&E staining of brain sections revealed extravasation of RBC in animals undergoing PIFS, but not in E7-treated control animals. In particular, extravascular erythrocytes were observed in the hypothalamus, cortex, hippocampus, striatum, and corpus callosum from moribund mice (Fig. 3, C and D). These experiments demonstrate that i.v. VP2121–130 peptide treatment induced blood vessel permeability in the brain.
In a separate experiment, we injected i.v. superparamagnetic Db:VP2121–130 tetramer into 7-day TMEV-infected animals to determine whether brain-infiltrating Db:VP2121–130 epitope-specific CD8+ T cells could be visualized using MRI. Injection of i.v. superparamagnetic Db:VP2121–130 tetramer induced PIFS in these animals. Moreover, MRI revealed extensive abnormalities in the CNS of tetramer-treated animals as demonstrated by T2-weighted images. As shown in Fig. 3F, the corpus callosum, hippocampus, cortex, thalamus, and striatum show evidence of tissue abnormality (normal tissue is dark in this analysis, whereas inflamed areas of tissue are light or washed out) 1 day following the administration of superparamagnetic Db:VP2121–130 tetramer. No such abnormality was observed in animals administered an irrelevant superparamagnetic Db:E7 tetramer (Fig. 3E). This demonstrated that, in addition to vessel permeability, there is extensive damage to CNS tissue following the induction of VP2121–130 PIFS with superparamagnetic Db:VP2121–130 tetramer, suggesting that such technology to image specific lymphocytes could be potentially hazardous to use clinically.
Our findings of tissue damage in the brain by MRI and gross histological analysis is further supported by the activation of glia following systemic treatment with antigenic peptide. Up-regulated expression of glial fibrillary acidic protein, a pattern associated with the response of astrocytes to immunologic and traumatic injury, indicates that systemic peptide treatment leads to a widespread stress response in brain cells (Fig. 4) (34, 35). Tissue damage in the brain is normally associated with the influx of macrophages and the activation of resident microglial cells. The widespread appearance of cells bearing the macrophage/microglia marker F4/80 in the brains of C57BL/6 mice with PIFS is a marker for the onset of these inflammatory changes associated with damage in the brain. In contrast, TMEV-infected 129 mice treated systemically with peptide showed very little and localized evidence of macrophage infiltration or microglial activation (compare top and bottom right panels) and essentially no stress response by astrocytes (compare top and bottom left panels). This is further evidence that there are strong strain-dependent injury responses in the brain to systemic peptide treatment of mice undergoing an active T cell response.
Because perforin–/– mice were generated on the 129 genetic background, it remained possible that 129 genes carried from the original perforin–/– founder mouse could be responsible for resistance to VP2121–130 PIFS. To rule out background genes playing a role, we adoptively transferred perforin-competent GFP-expressing splenocytes into irradiated perforin–/– mice 1 day before TMEV infection. Following transfer, >50% of brain-infiltrating CD8+ T cells during acute TMEV infection expressed GFP and hence also expressed perforin (data not shown). Perforin–/– mice reconstituted with GFP and perforin-expressing splenocytes readily succumbed to VP2121–130 PIFS, with seven of eight animals becoming symptomatic or moribund within 24 h with systematic peptide treatment (Table V). Our analysis of the animals revealed that both VP2121–130 and E7 peptide-treated animals had GFP-expressing cells in the cortex, striatum, hypothalamus, corpus callosum, and hippocampus (Fig. 5, E and F; data not shown). This experiment demonstrates that VP2121–130 PIFS is dependent on the expression of perforin by CNS-infiltrating immune cells and rules out the possibility that the 129 genes conferring resistance to this syndrome influence perforin release by the peptide-specific T cells.
Discussion
These experiments demonstrate that i.v. administration of class I-restricted peptides or class I tetramers that contain these peptides can be fatal to individuals harboring expanded, peptide-specific CD8+ populations. There is one well-characterized mechanism by which a response can lead to rapid death. Pathogen-derived superantigens simultaneously bind both the receptor (TCR) V chain and class II molecules (36). This cross-linking of TCR with class II leads to the polyclonal activation of CD4+ Th cells followed by the systemic increase of cytokines. Shock results from a dramatic rise in the cytokines TNF-, TNF-, IL-1, IL-6, IL-8, and IL-10. These cytokines initiate a cascade of fever, increased cardiac output, increased blood clotting, organ failure, and death (31). Our analysis rules out these classical shock pathways. In this study, we address a second T cell-mediated syndrome that is initiated by peptide-specific CD8+ T cells. Although CD4+ T cell-mediated shock is becoming better understood, the possibility of CD8+ T cells mediating similar syndromes has gone almost completely unnoticed. One study describing the possibility of CD8+ T cell-mediated shock used Abs that initiate polyclonal activation and their secretion of IFN- and TNF-. These cytokines were protective in the host response to Trypanosoma cruzi. However, when these T. cruzi-infected mice were administered anti-CD3 Ab, these mice underwent septic shock. Pretreatment with anti-CD8 Abs before this anti-CD3 treatment prevented septic shock (37). These observations suggested that, in this disease model, septic shock might be mediated by CD8+ T cells.
The importance of CD8+ T cells in inducing shock was suggested in other studies involving LCMV and tumor peptide vaccination (33, 38, 39). In the LCMV model, administration of peptide activated peptide-specific memory cells, inducing damage in the spleen that was generally immunosuppressive (38). Shock induced with LCMV was mediated by lymphotoxin as recombinant soluble chimeric lymphotoxin -receptor/Ab (LT-R) blocked virus-induced pathology (33). In the tumor peptide vaccination model, CD8+ T cell-mediated shock was mediated by TNF because anti-TNF Abs neutralized this fatal condition (39). In this study, we demonstrate that a fatal syndrome can be induced by the treatment of virus-infected mice with soluble peptide. Mice succumbing to PIFS display a dramatic breakdown of vascular integrity in the brain. In our system, TNF-, TNFRI, and TNFRII are not required for the development of the PIFS. Furthermore, induction of PIFS also is not influenced by treating animals with soluble LT-R fusion protein, indicating that none of the previously defined cytokine pathways appears to play a dominant role in this syndrome. Another possibility is that CD8+ T cells secrete cytokines that directly or indirectly lead to vessel permeability. Vessel permeability caused by cytokines has been reported by those investigating vascular leak syndrome (VLS), a condition where vascular epithelial cells become damaged and permeable to vascular fluids (40). In particular, the administration of rIL-2, a commonly emitted CD8+ cytokine, can initiate murine VLS (40, 41, 42). Although VLS is a trait of VP2121–130 PIFS, permeability of blood vessels in the brain alone does not lead to death. Perforin-deficient mice used in this study have extensive vessel permeability in the CNS comparable to C57BL/6 controls, but remain asymptomatic. Therefore, we hypothesize that perforin destabilizes cells in the CNS involved with vital functions. This is consistent with previous observations that perforin contributes to pathogenesis (28, 29, 43, 44, 45). Therefore, we may be triggering a CD8+ T cell-mediated process that occurs naturally following infection in the CNS.
The observation that CD8+ T cells can mediate a PIFS extends beyond the Theiler’s virus system. We have investigated whether peptide could induce shock in the 2C transgenic mouse. 2C transgenic mice express receptors that are reactive with the SIYR peptide presented in the context of Kb class I molecules on >90% of peripheral T cells. The effect of injecting SIYR peptide i.v. into healthy 2C TCR transgenic mice was similar to that observed upon VP2121–130 peptide treatment of 8-day TMEV-infected C57BL/6, class II–/–, and IFN-R–/– mice. PIFS was not reproduced with an irrelevant Kb-binding OVA peptide, demonstrating the peptide specificity of the syndrome. As little as 0.001 mg of SIYR peptide administered three times i.v. could induce PIFS in 2C transgenic mice. Similar to the VP2121–130 peptide-induced syndrome, SIYR PIFS was mediated by an expanded population of epitope-specific CD8+ T cells.
Peptide-specific CD8+ T cells are necessary, but not sufficient, to render animals susceptible to PIFS. Besides expression of perforin, PIFS is also influenced by the genetic makeup of the host. More specifically, we present evidence that resistance to PIFS is conferred through a complex interaction of multiple genes. The influence of mouse genotype can, therefore, explain how 129 mice containing populations of Db:VP2121–130 epitope-specific CD8+ T cells equivalent to C57BL/6 mice can remain resistant to VP2121–130 PIFS. More work is needed to determine whether mouse genotype influences perforin function. Because it is likely that peptide-mediated suppression will continue to be a strategy to down-regulate subsets of Ag-specific T cells, we find these observations to be highly relevant to those interested in the development of peptide therapies.
Acknowledgments
We thank John Altman of Emory University (Atlanta, GA) and Mark Davis of Stanford University (Stanford, CA) for contributing the Db and human 2m expression plasmids. We also thank Lieping Chen of Mayo for soluble LT-R. We also thank Mike Bell, Charles Howe, Kevin Pavelko, Loc Nguyen, and Yanice Mendez-Fernandez for technical assistance in the completion of this work.
Footnotes
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.
1 This work was funded by National Institutes of Health Grants N524180 and N532129.
2 Address correspondence and reprint requests to Dr. Larry R. Pease, Department of Immunology, Mayo Clinic College of Medicine, 200 First Street SW, Rochester, MN 55905. E-mail address: pease.larry{at}mayo.edu
3 Abbreviations used in this paper: TMEV, Theiler’s murine encephalomyelitis virus; PIFS, peptide-induced fatal syndrome; MRI, magnetic resonance imaging; LCMV, lymphocytic choriomeningitis virus; LT, lymphotoxin ; VLS, vascular leak syndrome.
Received for publication November 17, 2004. Accepted for publication February 18, 2005.
References
Hoyne, G. F., R. E. O’Hehir, D. C. Wraith, W. R. Thomas, J. R. Lamb. 1993. Inhibition of and antibody responses to house dust mite allergen by inhalation of the dominant epitope in naive and sensitized mice. J. Exp. Med. 178: 1783-1788.
Rolland, J., R. O’Hehih. 1998. Immunotherapy of allergy: anergy, deletion, and immune deviation. Curr. Opin. Immunol. 10: 640-645.
Muller, U., C. A. Akdis, M. Fricker, M. Akdis, T. Blesken, F. Bettens, K. Blaser. 1998. Successful immunotherapy with T-cell epitope peptides of bee venom phospholipase A2 induces specific T-cell anergy in patients allergic to bee venom. J. Allergy Clin. Immunol. 101: 747-754.
Magee, C. C., M. H. Sayegh. 1998. Peptide mediated immunosuppression: new developments. Transplant. Proc. 30: 2131-2135.
Magee, C. C., M. H. Sayegh. 1997. Peptide-mediated immunosuppression. Curr. Opin. Immunol. 9: 669-675.
Minev, B. R., F. L. Chavez, M. S. Mitchell. 1998. New trends in development of cancer vaccines. In Vivo 12: 629-638.(Aaron J. Johnson*,, Yanic)