transplantation
Oosten, L.E.M.
Citation
Oosten, L. E. M. (2007, November 21). Adoptive immunotherapy after HLA
mismatched stem cell transplantation. Retrieved from
https://hdl.handle.net/1887/12446
Version: Corrected Publisher’s Version
License: Licence agreement concerning inclusion of doctoral
thesis in the Institutional Repository of the University
of Leiden
Downloaded from: https://hdl.handle.net/1887/12446
Note: To cite this publication please use the final published version (if
applicable).
101
General discussion
Chapter 6:
103 1. ALLOHLA-RESTRICTED T CELLS AS REAGENTS FOR ADOPTIVE IMMUNO-
THERAPY AFTER HLA-MISMATCHED STEM CELL TRANSPLANTATION
Stem cell transplantation (SCT) is a generally accepted treatment for patients with hematological malignancies. Stem cell donor-derived T cells mediate both benefi cial graft-versus-leukemia (GvL) effects and detrimental graft-versus-host disease (GvHD) after allogeneic SCT. Relapse of the original disease can be treated with infusions of donor lymphocytes (DLI). Like SCT, DLI elicits both GvHD and GvL. The risk of severe GvHD is especially high if the stem cell donor and patient are not fully matched for human leukocyte antigens (HLA). To minimize GvHD after HLA-mismatched SCT DLI should consist of pre-selected donor T cells that display reactivity restricted to the patient’s residual leukemic or hematopoietic cells. The use of ex vivo generated T cells that selectively recognize such antigens would thus greatly increase the number of patients eligible for immunotherapy in the HLA-mismatched setting. This thesis focuses on exploring the alloHLA-A2-reactive T cell repertoire and testing the feasibility of generating alloHLA-A2-restricted T cells specifi c for the hematopoietic system-restricted minor histocompatibility antigens (mHags) HA-1 or HA-2.
2. GENERATION OF MHAG-SPECIFIC ALLOHLA-RESTRICTED T CELLS:
EXPLORING MODIFIED ANTIGEN-PRESENTING CELLS
2.1 STIMULATION OF MHAG-SPECIFIC T CELLS USING ALTERNATIVE ANTIGEN- PRESENTING CELLS
Natural antigen-presenting cells (APCs) express multiple different HLA/peptide complexes on their cell surface. Currently available protocols for the ex vivo generation and expansion of alloHLA-restricted T cells using natural APCs results in adventitious co-stimulation of broad and unpredictable alloHLA-reactivities. Such alloHLA-reactivities could cause GvHD after HLA-mismatched SCT. To circumvent this problem we aimed to develop APCs that present only a selected set of functional HLA/peptide ligands to stimulate mHag-specifi c T cells. In chapter 2, we explored the use of artifi cial antigen-presenting constructs (aAPCs) coated with a single type of HLA-A2/mHag complex. In chapter 3, we examined the effects of blocking the transporter associated with antigen processing (TAP) on the ability of natural APCs to stimulate alloHLA-A2-reactive T cells.
2.2 STIMULATION OF MHAG-SPECIFIC T CELLS BY ARTIFICIAL ANTIGEN-PRESENTING CONSTRUCTS
In chapter 2, we showed that aAPCs consisting of latex beads coated with HLA-A2/mHag complexes and the costimulatory molecules, CD80 and CD54 effi ciently stimulate mHag- specifi c T cell clones in a ligand-density dependent manner and can be used to enrich polyclonal mHag-specifi c T cell lines without affecting T cell phenotype or cytolytic
104
activity. Furthermore, these aAPCs do not stimulate broad alloHLA-A2-reactive clonal or polyclonal T cells (chapter 4). HLA-A2/mHag-coated aAPCs might thus serve as tools for in vitro expansion of mHag-specifi c alloHLA-restricted T cells present in a pool of alloHLA-reactive T cells.
Currently, the majority of protocols for in vitro expansion of immunotherapeutic T cells involve stimulation with autologous dendritic cells (DCs). The generation of large numbers of monocyte-derived DCs is laborious and time-consuming. Bead-based aAPCs pose an easy-to-use substitute, amenable to good manufacturing practices (GMP). Furthermore, the number and nature of stimulatory ligands on the surface of bead-based aAPCs can be precisely controlled.
Several studies have assessed bead-based aAPCs as agents for primary in vitro induction of antigen-specific T cells. Cytomegalovirus (CMV)-specific and tumor-associated Melan-A/MART-1 antigen-specifi c CD8+ T cells were generated from peripheral blood mononuclear cells (PBMCs) after stimulation with bead-based aAPCs1,2. We could also effi ciently generate CMV pp65-specifi c T cells from PBMCs of HLA-A2pos individuals using HLA-A2/pp65 aAPCs (unpublished data). However, we could not induce mHag-specifi c T cells with HLA-A2/mHag aAPCs, neither from HLA-A2pos nor from HLA-A2neg donors (unpublished observations). Several of these donors had earlier yielded mHag-specifi c T cells using a conventional stimulation protocol including natural APCs. We therefore assume that our aAPCs do not provide adequate stimulation to activate truly naive T cells. T cell priming in vivo is tightly regulated and requires highly professional APCs to provide a precise mixture of cytokines and costimulatory signals. Because most individuals have been primed for CMV and because activated Melan-A/MART-1-specifi c T cells can be detected in the peripheral blood of healthy HLA-A2pos donors3, aAPC-mediated generation of T cells specifi c for these antigens may be due to ex vivo expansion rather than de novo induction. A recently published study by Schilbach et al. indicated that primary in vitro induction of HA-1-specifi c T cells could be achieved with aAPCs coated with HLA-A2/
HA-1 complexes and anti-CD28 antibodies4. Further optimization of HLA-A2/mHag complex-coated aAPCs exploited and discussed below may yield more effi cient T cell induction/expansion protocols for adoptive immunotherapy with mHag-specifi c T cells after HLA-A2-matched as well as HLA-A2-mismatched SCT.
2.2.1 OPTIMIZING ARTIFICIAL ANTIGEN PRESENTATION
The fi rst requirement for effective T cell stimulation by aAPCs is adequate expression of HLA/peptide complexes. Attaching HLA/peptide complexes to the bead surface via streptavidin or an antibody-linker ensures appropriate presentation of the complex to the T cell receptor (TCR) and increases fl exibility of the interaction5. Because the rigid surface of aAPCs impairs lateral movement of ligands, a degree of fl exibility is required to allow for TCR clustering on the T cell surface, an essential step in T cell activation6. Furthermore, the density of HLA/peptide complexes expressed on aAPCs strongly affects T cell avidity. Low-density aAPCs have been shown to induce T cells with higher avidity than
105 high-density aAPCs1,4. We optimized our aAPCs for the effi cient stimulation of previously
established mHag-specifi c T cell clones. Using lower ligand densities may improve the success rate of aAPC-mediated T cell priming.
Virtually all bead-based aAPCs provide a ligand for the costimulatory CD28 receptor present on T cells. This can be CD80 (B7-1), as exploited in our system, or an agonistic anti-CD28 antibody. T cell stimulation by CD80 more closely resembles the physiological interaction between T cells and APC, but can also trigger the inhibitory receptor CD152 (CTLA-4) expressed by recently activated T cells or by regulatory T cells present in the same culture7. The use of anti-CD28 antibodies has proven successful for in vitro induction of HA-1-specifi c T cells4 and should perhaps have preference. Like the aAPCs described in this thesis, many aAPCs are additionally coated with CD54 (ICAM-1) to promote adhesion. Its synergistic effect with CD28 has been extensively documented8,9, and has been corroborated by our own fi ndings. Furthermore, we investigated the addition of the accessory molecule CD137L (4-1BBL). Ligation of CD137L with CD137 on activated T cells is important in the induction, amplifi cation, and persistence of murine T cells10,11 and is associated with reduced apoptosis of recently primed human CD8+ T cells12. We coated our aAPCs with varying concentrations of CD137L, but could not detect an appreciable effect on the survival of mHag-specifi c T cell clones (unpublished data). Whether the addition of CD137L facilitates in vitro induction of mHag-specifi c T cells remains to be seen. Another candidate molecule for coupling to aAPCs is CD7013. In mice, CD27-CD70 interactions seem to be important for long-term CD4+ and CD8+ T cell survival and/or expansion of memory T cells13. Since the various costimulatory pathways mentioned above seem to work at different time points early in the process of T cell priming and differentiation, it will be diffi cult to determine the most optimal ligand combination on aAPCs.
Cytokines are additional critical determinants of T cell expansion, survival, and function.
Mature DCs produce several cytokines including interleukin-2 (IL-2), IL-7, IL-12, and IL- 15. Zeng et al. recently demonstrated that IL-21, primarily together with IL-15 or IL-7, potently regulates CD8+ T cell expansion and effector function in mice14. We found that addition of T cell growth factor (T-CGF) instead of IL-2 greatly improved T cell expansion in our system (unpublished data). T-CGF, which consists of the supernatants of mitogen- stimulated PBMCs, is unfortunately not available at GMP grade. In the study by Schilbach et al.IL-15 was added in addition to IL-2, IL-7 and IL-12, with apparent success4. Thus, optimizing the recombinant cytokine mixture in combination with improved aAPCs may be of great importance for the effi cient and reproducible in vitro induction/expansion of mHag-specifi c T cells.
2.3 REDUCTION OF ALLOHLA-REACTIVITY BY INHIBITION OF TAP
TAP is a key molecule in the translocation and HLA class I loading of (allo) peptides. We hypothesized that retroviral transduction of APCs with a viral TAP inhibitor would reduce their alloHLA antigenicity. Allogeneic HLA-A2pos APCs, transduced with a TAP inhibitor and exogenously pulsed with mHag peptide, could then be used to generate mHag-specifi c
106
T cells from HLA-A2neg donors; these T cells could subsequently be expanded with aAPCs.
In chapter 3, we compared three different TAP inhibitors i.e. US6, ICP47 or UL49.5, derived from human CMV, herpes simplexvirus type 1, and bovine herpes virus) type 1.
Retrovirally transduced Epstein-Barr virus-transformed lymphoblastoid cell lines (EBV- LCLs) exhibited a stable decrease in cell surface HLA class I expression and were protected from lysis by mHag-specifi c T cells. Exogenous addition of mHag peptide fully restored target cell recognition. UL49.5 was the most potent inhibitor of TAP, reducing cell surface HLA class I expression and mHag-specifi c lysis to the level of TAP-defi cient cell line T2.
While UL49.5 inhibited alloHLA-A2- and alloHLA-A1-specifi c lysis by alloHLA-reactive T cell clones, reduction of lysis was incomplete. Thus, an APC that expresses UL49.5 and has been exogenously pulsed with mHag peptides may still induce undesired alloHLA- reactive T cells. While UL49.5 downregulated alloHLA-A2 recognition more effectively than ICP47, increased presentation of signal-sequence derived peptides associated with TAP inhibition15 was observed for ICP47 only (personal communication, T van Hall).
Apparently, UL49.5 inhibits other constituents of the major histocompatibility complex (MHC) loading complex, most likely tapasin16, as well. These fi ndings indicate that further exploration of UL49.5 as a tool to modulate endogenous antigen presentation by APCs is justifi ed. However, the residual alloHLA-reactivity directed against TAP-defi cient cell line T2 due to continued expression of TAP-independent peptides at the cell surface is considerable. Inhibition of TAP alone may therefore prove insuffi cient to reduce an APC’s alloHLA antigenicity to the level required for clinical application.
3. SPECIFICITY OF MHAG-SPECIFIC ALLOHLA-RESTRICTED T CELLS:
EXPLORING THE ALLOHLA REPERTOIRE
3.1 EVALUATION OF THE ALLOHLA-A2-SPECIFIC T CELL REPERTOIRE
Two models have been proposed to account for alloHLA-reactivity: the high determinant density model suggesting that alloHLA-reactive T cells recognize alloHLA molecules independent of bound peptide, and the multiple binary complex hypothesis stating that alloHLA-reactive T cells respond to a combination of alloHLA-molecule and peptide.
Peptide-independent and peptide-specifi c alloHLA-recognition have both been reported17-
23. For the generation of mHag-specifi c allo-HLA-restricted T cells to be feasible and reproducible, the second model must take precedence.
In chapter 4, we investigated alloHLA-A2-reactive T cell responses in various HLA-A2neg donors using tetrameric HLA-A2/peptide complexes.Tetramers have been shown to bind to antigen-specifi c T cells with high specifi city and tetramer binding often correlates with peptide-specifi c cytolytic and cytokine-secreting functions24,25. Using HLA-A2pos EBV-LCLs or HLA-A2pos TAP-defi cient T2 cells as stimulator cells, we detected, within alloHLA-A2- reactive T cell populations, distinct subsets of CD8+ T cells specifi c for peptides derived from ubiquitously expressed self-proteins known to be presented in the context of HLA-
107 A2. None of these alloHLA-A2-reactive T cell populations could be stimulated by aAPCs
coated with single sets of HLA-A2/ irrelevant peptide complexes. Thus, the majority of alloHLA-A2-reactive T cell lines, generated using a conventional stimulation protocol, displayed a peptide-specifi c recognition pattern.
These fi ndings elaborate upon earlier publications on alloHLA-recognition. One study reported that 25-80% of alloHLA-A2-reactive T cells is peptide-independent, as defi ned by lytic responses to unpulsed T2 cells21. Using HLA/peptide tetramers we were able to show that alloHLA-A2-restricted T cells can be specifi c for signal sequence-derived peptides presented by T2 and that the majority of alloHLA-A2-restricted T cells do not recognize HLA-A2 independent of bound peptide. Our data therefore allow us to exclude peptide- independent alloHLA-A2 recognition as a major contributor to the alloHLA-A2-response.
However, the presence of peptide-dependent but not peptide-specifi c alloHLA-A2-restricted T cells cannot be excluded.
3.2 GENERATION OF ALLOHLA-REACTIVE T CELLS USING COMPLEX-COATED DCS
Considering the potent stimulatory capacity of HLA-A2/mHag complexes (chapter 2) and their failure to stimulate polyclonal alloHLA-A2-reactive T cell populations (chapter 4), we explored the use of mature HLA-A2neg DCs coated with trimeric HLA-A2/HA-1 complexes as HA-1-specifi c stimulators of autologous HLA-A2neg T cells in chapter 5. The feasibility of generating high avidity alloHLA-A2-restricted HA-1-specifi c T cells from certain HLA-A2neg male donors using HA-1 peptide-pulsed T2 cells or allogeneic HLA-A2pos DCs retrovirally transduced with HA-1 was previously established26. We generated mature autologous DCs from the same two HLA-A2neg donors and coated these cells with recombinant HLA- A2/HA-1 complexes. The coating protocol was optimized for maximum stimulation of previously established mHag-specifi c T cell clones. Using HLA-A2/HA-1 complex-coated DCs as stimulator cells, we could induce small populations of alloHLA-A2-restricted T cells that bound HLA-A2/HA-1 tetramers from the PBMCs of both donors. In contrast to our fi ndings in the HLA-A2-matched setting27, tetramer binding did not correlate with lytic function. T cells generated from one donor selectively bound HLA-A2/HA-1 tetramers and displayed specifi c lysis of T2 cells pulsed with HA-1 peptide. Yet, cytotoxicity profi les revealed that all HLA-A2pos target cells were lysed irrespective of the presence of HA-1. T cells generated from the other donor did not bind HLA-A2/HA-1 tetramers but did display specifi c HA-1-specifi c lysis at high ligand-densities, probably due to low affi nity TCRs. Thus, neither donor exclusively yielded high avidity T cells that were truly specifi c for HA-1.
While these data could refl ect characteristics of the alloHLA-A2-restricted T cell repertoire, they could also result from our experimental set-up using DCs coated with trimeric HLA- A2/HA-1 complexes via anti-HLA-DR antibodies as APCs. This set-up was optimized for stimulation of previously established mHag-specifi c T cell clones, but may have been suboptimal for activation of naive mHag-specifi c T cells. One explanation is that the
“elongated” HLA-A2/HA-1 ligand may have reduced TCR triggering without affecting TCR-ligand ligation28 or affected CD8 coreceptor binding (“signal 1”). In addition,
108
the “elongated” HLA-A2/HA-1 ligand may have affected the T cell’s ability to span the distance between its own and the opposing membrane of the DC, resulting in insuffi cient interaction between costimulatory molecules on the T cell and their ligands on the DC (“signal 2”). Both signal 1 and signal 2 are required for optimal T cell priming resulting in high avidity T cells. An earlier study using bead-based aAPCs showed that indirect conjugation of HLA/peptide ligand via anti-HLA-DR-antibodies resulted in more effi cient stimulation of T cells than direct conjugation of ligand to the bead surface5. However, this study addressed expansion of T cells only, not in vitro priming. Also, agonistic CD28 antibody was used as a costimulatory molecule which has a different physical dimension than its ligand CD80 that is naturally present on activated DCs including our own.
A second handicap of our experimental set-up may have been the potent stimulatory capacity of HLA-A2/HA-1 complex-coated DCs. The use of such highly professional APCs may have inadvertently facilitated expansion of low avidity crossreactive memory T cells naturally present in the alloHLA-A2-restricted T cell repertoire29. Once activated, these non-specifi c T cells are likely to compete with the de novo generated HA-1-specifi c T cells for available growth factors.
Furthermore, the high density of HLA-A2/HA-1 ligand on the APCs may have skewed the induced T cell populations to T cells with low avidity only. This would be in line with the previously mentioned report on the infl uence of ligand density on aAPC-mediated T cell stimulation1,4. Yet, alloHLA-restricted T cells displaying similar avidity profi les have been obtained with other more conventional stimulatory protocols20,23,30,31. Thus, the induction of high avidity (tumor) antigen-specifi c T cells may be intrinsically more diffi cult in the HLA-A2-mismatched setting compared to the HLA-A2-matched setting.
3.3 DIRECT ISOLATION OF ALLOHLA-REACTIVE T CELLS FROM PERIPHERAL BLOOD In the second approach described in chapter 5, we attempted to directly isolate alloHLA- A2-restricted HA-2-specifi c T cells from peripheral blood samples of HLA-A2neg female donors naturally exposed to HLA-A2 through previous pregnancies. Pregnancy can immunize for paternal alloHLA molecules and mHags expressed by the fetus32-34. CD8+ HLA-A2/HA-2 tetramer-binding T cells were isolated from CD8+-enriched PBMC fractions and non-specifi cally expanded in the presence of autologous feeder cells. After 28 days of in vitro culture, the T cell populations obtained from two donors displayed HLA-A2/
HA-2 tetramer-binding T cells. One polyclonal T cell line selectively bound HLA-A2/
HA-2 tetramers but was shown to contain T cells of low avidity only upon functional testing. The other T cell population contained a T cell subset that bound three out of fi ve HLA-A2/peptide tetramers, indicating peptide-selective but not peptide-specifi c alloantigen recognition. We examined the alloHLA-A2-reactive T cell repertoire of both donors in more detail by direct ex vivo cloning of tetramer-binding T cells. The resulting T cell clones either bound none of the HLA-A2/peptide tetramers tested, several of the tetramers tested, or HLA-A2/HA-2 tetramers only. None of the T cell clones displayed HA-2-specifi c lysis in a cytotoxicity assay. These data show that HLA-A2 tetramers are
109 unreliable tools for isolation of truly HA-2-specifi c T cells from the promiscuous alloHLA-
A2-specifi c repertoire.
The protocol applied has previously been used to successfully isolate HA-1- or HA- 2-specific cytotoxic T cells from the peripheral blood of HLA-A2pos mHagneg female donors who had delivered HLA-A2pos mHagposchildren32. However, exposure to fetal alloantigens during pregnancy is likely to induce regulatory T cells that may protect the fetus from potentially harmful alloresponses by the mother35,36. The putative presence of regulatory T cells might affect the overall low functionality of the HA-2-specifi c T cell lines obtained from the fi rst donor. After we sub-cloned this T cell population, a pool of poorly proliferating HLA-A2/HA-2 tetramer-binding T cell lines produced not only IFN-γ but also IL-10 in response to specifi c stimulation (unpublished data), which is a feature of peripherally induced regulatory T cells37. “Single cell per well” tetramer-based fl uorescence-activated cell sorter (FACS) sorting allowed us to isolate several high avidity alloHLA-A2-reactive T cell clones from the same donor. Interestingly, one of the isolated T cell clones bound all HLA-A2 tetramers tested with high intensity but showed no lytic activity to HLA-A2pos target cells whatsoever. Further studies may reveal whether we have isolated another low avidity T cell or whether we have isolated the fi rst broad alloHLA-A2 reactive regulatory T cell.
In summary, in vivo induced alloHLA-A2-restricted mHag-specific T cells share characteristics of peptide selectivity but not peptide specifi city with alloHLA-A2-restricted T cells generated in vitro using HLA-A2/mHag complex-coated DCs. In the latter case these characteristics could theoretically be due to our experimental set-up. In the case of in vivo induced T cells this is unlikely as the T cells were expanded ex vivo using cytokines and phytohemagglutanin only. Interestingly, a recent study on direct isolation of alloHLA- A2/Melan-A-specifi c T cells from PBMCs using tetramers yielded T cells with similar characteristics38. Together, these results strongly indicate that the alloHLA-A2-restricted repertoire contains a broad spectrum of promiscuous T cells.
4. CROSSREACTIVITY AND ALLOREACTIVITY
4.1 RECENT INSIGHTS INTO CROSSREACTIVITY OF T CELL RECOGNITION
Over the last decade, notions on the nature of TCR – HLA/peptide interactions have changed considerably. This interaction was originally thought to be highly specifi c as a TCR can discriminate between peptides differing in a single amino acid. However, evidence is emerging that TCR crossreactivity is common and represents an important aspect of TCR recognition.
During T cell development, immature T cells are selected to mature only if their TCRs are capable of interacting with self-HLA/self-peptide complexes expressed in the thymus (positive selection). It is now known that the long-term survival of mature T cells in the periphery also depends on low affi nity interactions between their TCRs and self-HLA/peptide complexes39.
110
These interactions seem to be peptide-specifi c. Homeostatic proliferation of murine CD8+ T cells expressing transgenic TCRs is diminished in TAP-defi cient hosts, and can be restored by the addition of a MHC class I-binding peptide known to form a low affi nity ligand for the TCR. A control peptide binding the same MHC class I molecule had no such effect40. TCRs must therefore be able to bind at least one self-HLA/self-peptide complex in addition to any self-HLA/foreign peptide complexes. To enable the available T cell repertoire to recognize the vast number of potential HLA/peptide complexes presented in an individual, T cells must in fact possess much more specifi cities. Theoretical analysis suggests that a single TCR should be capable of recognizing 106 different peptide ligands to provide a T cell repertoire suffi ciently diverse to be functional41.A large number of studies have now demonstrated TCR crossreactivity for a variety of human CD8+ T cells (reviewed42,43).
A well defi ned example is the melanoma-associated antigen Melan-A. In healthy human HLA-A2pos donors up to 0.1% of circulating naive CD8+ T cells bind Melan-A tetramers.
The majority of these T cells display a very low functional avidity for Melan-A peptides and have been shown to recognize peptides derived from self-proteins selected for potential crossreactivity with Melan-A3. Interestingly, alloHLA-A2/Melan-A-specifi c T cells displaying the same crossreactivity profi le have been isolated from HLA-A2neg donors44. Both “auto”- and alloHLA-A2/Melan-A2-specific T cells predominantly express the Vα2.1 chain combined with a variety of Vβ chains38. The existence of such a degenerate recognition of Melan-A-related epitopes in both the autologous and allogeneous setting by different but structurally related TCRs indicates that a considerable number of self-peptides could contribute to the repertoire selection of Melan-A2 tetramer-binding T cells.
4.1.1 MECHANISMS OF CROSSREACTIVITY
Crystallographic analyses of TCR structures have helped to elucidate the mechanisms supporting crossreactivity (reviewed45). The peptide-binding CDR3 loops of the TCR are highly fl exible in the unbound state and may rearrange upon the binding of HLA/peptide complexes to accommodate differences in peptide sequence46,47. In contrast, the CDR1 and CDR2 loops that mainly interact with the α helices of the HLA molecule are more rigid and show little or no rearrangement upon HLA/peptide complex binding. Based on these notions, a two-step mechanism for TCR recognition has been proposed48. First, CDR1/2-HLA interactions facilitate initial contact between the TCR and its ligand. Then, the CDR3 loops fold over the peptide and establish defi nite binding. This two-step model allows for the scanning of the high number of peptides displayed by HLA molecules on APCs. TCRs may sample HLA/peptide complexes in a way that is mainly independent of bound peptide, whereas T cell activation is only triggered upon the formation of a stable, high affi nity contact between CDR3 and peptide. Because the fl exibility of the CDR3 loops enables “induced fi t” with different peptides, crossreactivity may ensue.
The spectrum of peptides to which a given T cell clone can react is considerable (reviewed49).
Obviously, crossreactivity can be based on sequence homology, but crossreactivity with peptides displaying minimal sequence similarity has been reported as well50,51.
111 Crystallography revealed that some amino acid residues serve as contact residues for the
TCR while others facilitate binding to the HLA molecule. One study in a murine system showed that conservation of three out of nine residues was suffi cient to maintain T cell recognition52. TCRs may even bind ligands unrelated in amino acid sequence53, indicating that three dimensional structure rather than amino acid sequence determines T cell recognition. No inherent differences exist between the TCRs of CD8+ and CD4+ T cells. Yet, CD8+ T cells seem to be somewhat less crossreactive than CD4+ T cells. This is probably due to the less constrained binding of peptides within the HLA class II peptide-binding groove. It has been suggested that TCRs primarily recognizing the N-terminal portion of peptides bound to HLA class II molecules are more inherently crossreactive than TCRs targeting the central portion since the conformation of peptides binding the central residues is more conserved54. Another reason for the reduced crossreactivity of CD8+ T cells as compared to CD4+ T cells could be the generally higher affi nity interactions between CD8+ TCR and HLA/peptide complexes55. Nevertheless, one study showed that 71% of HLA-A2-restricted CD8+ T cells generated against a haptenated peptide could crossreact with at least one other hapten56.
A recently identifi ed crystal structure for an autoimmune TCR bound to an HLA/peptide complex literally added a new dimension to crossreactivity57. In the “conventional”
diagonal or orthogonal orientation of TCR - HLA/peptide complex interactions, the variable α-region of the TCR (Vα) is atop the α2 (class I) or β2 (class II) helix, whereas Vβ is atop the α2 helix. In the newly described docking mode, Vα and Vβ are on the same side of the HLA molecule, contacting only the N-terminal half of the peptide. This structure might be representative of a greater range of possible TCR orientations on HLA/peptide complexes than hitherto appreciated. If so, these fi ndings extend the scope of potential crossreactivities even further.
4.1.2 FINE-TUNING OF A CROSSREACTIVE T CELL REPERTOIRE
If crossreactivity is common, most TCRs will be capable of recognizing self-peptides as well as foreign peptides. To avoid autoimmunity, undesired self-reactivities must be kept in check. The immune system employs several strategies to inhibit the expansion of low avidity self-reactive T cells.
Experiments with altered peptide ligands (APLs) have revealed that T cell recognition is not an all-or-non phenomenon58. The same T cell can be fully activated, partially activated as defi ned by differences in effector function, inhibited, or remain unaffected by APLs differing slightly in TCR contact residues. T cell avidity for a peptide is determined primarily by the affi nity of its TCR for the resulting HLA/peptide complex. The expression of co-receptors and signaling molecules further affects its avidity. A T cell will respond to a low ligand density if its avidity is high, but will require a higher ligand density if its avidity is low. Because a single TCR can bind different HLA/peptide complexes with different affi nities, responses of a crossreactive T cell to its potential ligands may differ considerably in quality and quantity.
112
During T cell development, immature T cells are deleted if they respond too strongly to self-HLA/self-peptide complexes expressed in the thymus (negative selection). Negative selection also removes T cells expressing very crossreactive TCRs as was recently shown in a murine system59. Yet, T cells displaying low avidity for a limited set of self-antigens may escape thymic deletion. Low avidity for self-antigens renders a T cell insensitive to physiological densities of self-ligands, but avidity for self-HLA/foreign peptide complexes may still be high.
In addition, individual T cells can be desensitized by exposure to low avidity self-ligands both in the thymus and in the periphery. Murine TCR-transgenic cells transferred into a lymphopenic recipient expressing their cognate antigen become increasingly insensitive to antigenic stimulation60. This adaptation requires continuous expression of antigen, as the transfer of desensitized cells to an antigenneg host returned their sensitivity to normal levels. “Tuning” of T cell avidity probably has a biochemical basis in fl uctuations in the levels of excitatory and inhibitory molecules within the cell (reviewed61).
The exposure of murine polyclonal antigen-specific T cell populations to varying concentrations of a single peptide has shown that high avidity T cells dominate at low antigen doses whereas low avidity T cells dominate at high antigen doses62. In consequence, the immune response is mediated by T cells with the lowest possible avidity, minimizing the risk of concomitant self-reactivity. Thus, the mechanisms of tuning and avidity-based selection together tolerize the T cell repertoire reactive to self-peptides by limiting a crossreactive T cells’ self-reactivity while maintaining the capacity to respond to self-HLA/
foreign peptide complexes.
Interestingly, when the alloHLA-A2-reactive T cells from the fi rst donor in chapter 5 were stimulated with HLA-A2pos EBV-LCLs naturally expressing HA-1, instead of HLA-A2neg HLA- A2/HA-1 complex-coated DCs, the T cells’ alloHLA-A2 reactivity increased signifi cantly.
After two rounds of stimulation T2 cells were lysed independent of the type of peptide used for exogenous peptide pulsing. In this case stimulating with APCs expressing lower levels of ligand may actually have increased the alloHLA-A2-reactive T cells’ avidity.
4.1.3 VIRAL INFLUENCES ON CROSSREACTIVITY OF THE T CELL REPERTOIRE
While the mechanisms described above effectively decrease crossreactivity of the T cell repertoire, infections with pathogens do the opposite. For example, infection of mice with lymphocytic choriomeningitis virus protected them from subsequent infections with pichinde virus and vaccinia virus63,64. In humans crossreactive T cell responses to different pathogens have been observed65,66. Because crossreactive memory T cells have an advantage over naive T cells due to their high frequency and activation state, encountering a crossreactive antigen may lead to preferential expansion of these memory T cells67. Indeed, exposure to two variant Infl uenza A peptides, selectively expanded murine crossreactive T cells recognizing both the parental and mutant peptide from polyclonal memory pools68. This makes evolutionary sense, as many viruses are highly mutagenic. An individual’s history of infections could thus alter the shape of the T
113 cell repertoire or the nature of the T cell response. For instance, immunity to infl uenza
shifted the usual Th2 response to a challenge with respiratory syncytial virus to a Th1 response in mice69.
Such environmental infl uences could well explain the diffi culties in generating alloHLA- A2-restricted antigen-specifi c T cells from the same donor at different time-points as observed by others30,70 and ourselves in chapter 5. T cells proliferating in response to alloHLA-antigen stimulation have been shown to display a memory phenotype even if the donors had not been immunized against these antigens71. In our own study on the presence of HA-2-specifi c T cells in the peripheral blood of HLA-A2neg female donors potentially primed for HA-2, 32 - 57% of CD8+ T cells that bound HLA-A2/HA-2 tetramer at the time of FACS-sorting expressed CD45RO, indicating a memory phenotype. As these T cells were not HA-2-specifi c, the priming antigen may have been any HLA-A2- associated peptide directly presented by fetal cells (including HA-2 itself) or even a non-HLA-A2 associated antigen71. When we performed control double FACS-sorting on PBMC fractions derived from a male HLA-A2neg donor without a known history of prior immunization to HLA-A2, we also detected a percentage of CD45RO+, HLA-A2/HA-2 tetramer-binding T cells comparable to that of the female donors described in our study (unpublished data). These fi ndings support the notion that the T cell repertoire contains high numbers of crossreactive T cells with overlapping specifi cities, thus fi lling “holes” in the T cell repertoire. Probably, the TCR has evolved to optimally balance specifi city and crossreactivity. This balance represents a compromise between the necessity to ensure functional recognition of a large number of possible pathogenic epitopes and the need to avoid harmful autoreactivities.
4.2 CROSSREACTIVITY AFTER ALLOGENEIC SCT
Under physiological conditions, the mechanisms described above are suffi cient to protect an individual from overt autoimmunity (reviewed61). After allogeneic SCT however, the mature donor T cells within the stem cell graft are exposed to the host’s disparate HLA/peptide complexes. The donor T cell repertoire has not been selected or tuned to ignore these ligands.
Furthermore, the ligands are encountered under severe proinfl ammatory conditions due to the preparatory conditioning regimen. Donor T cell immune responses are therefore likely to ensue. How then, does crossreactivity come into play after allogeneic SCT?
After HLA-matched SCT, T cell immune responses are directed against disparate mHags presented in the context of shared HLA alleles. The donor T cell repertoire perceives disparate mHags as non-self and will respond accordingly. However, the majority of peptides presented by the host’s APCs will be perceived as self, because identical HLA molecules bind the same peptide selection. Donor T cells have been instructed to ignore these peptides. Therefore, crossreactivity after HLA-matched SCT will be limited to peptides derived from polymorphic proteins such as mHags.
After HLA-mismatched SCT, cotransferred donor T cells are confronted with a whole new peptide repertoire complexed with the patient’s alloHLA molecules. Because each HLA
114
molecule has different peptide-binding requirements, an alloHLA molecule will present a different selection of self-peptides. The donor T cell repertoire may perceive each of these peptides as non-self. A single T cell clone can thus be crossreactive with several peptides, all “self ” to the patient. Furthermore, negative selection in the thymus normally ensures that TCR recognition focuses on the peptide component of the self-HLA/peptide complex.
That selective criterion is absent for TCR recognition of alloHLA/peptide complexes.
Indeed, in a murine system TCR affi nities for alloMHC/peptide complexes were higher than for self-MHC/peptide complexes72 and associated with increased degeneracy of ligand recognition59. In other words, after thymic selection the surviving T cells’ TCRs have less stringent binding requirements for interactions with alloHLA/peptide complexes than for interactions with self-HLA/peptide complexes as the former is not a selection criterion.
In addition, the alloHLA-reactive T cell repertoire is influenced by anti-viral T cell responses. T cells crossreactive with both self and viral antigens are thought to be deleted in the thymus or rendered tolerant, with exceptions to this “rule” being implicated in autoimmune disease. No such culling occurs of viral antigen-specifi c T cells crossreactive with alloHLA-antigens. Many T cell clones have been identified that display dual specifi city for viral antigens complexed with self-HLA and for alloHLA/peptide complexes (reviewed73,74). In a murine system, “heterologous” donor T cells with such dual specifi city were involved in the boosting of alloHLA-responses in an HLA-mismatched host during viral infection75. Recently, a similar observation was made in human lung transplant patients (personal communication, N.A. Mifsud). Viral infections have long been known to precipitate the rejection of allografts76, but they may also hinder the induction of tolerance. In an effort to mimic the immune history of humans in a tolerance model, mice that had been serially infected with non-persistent viral pathogens were rendered refractory to tolerance induction77.
In summary, T cell crossreactivity is of far greater importance in the HLA-mismatched setting than in the HLA-matched setting.
4.3 A NEW MODEL FOR ALLOHLA-RECOGNITION
Based on the data presented in chapters 3, 4, 5, and the recent insights outlined above, a new model could be proposed to account for the high frequency of alloHLA-reactive T cells. We have shown that alloHLA-A2-reactive T cells do not respond to HLA-A2 independent of bound peptide. We have also shown that in vivo activated alloHLA-A2- reactive T cell clones are not specifi c for a single HLA-A2/mHag combination. Based on these observations it could be hypothesized that the strength of the alloHLA-response neither stems from peptide-independent recognition of the alloHLA molecule itself (the high determinant density model), nor from peptide-specifi c recognition of a new alloHLA molecule-associated peptide repertoire (the multiple binary complex), but from the increased degeneracy of alloHLA/peptide complex recognition (Figure 1). All recent human studies focusing on the generation of alloHLA-restricted peptide-specifi c T cells using in vitro tissue culture systems have reported the occurrence of low avidity T cells
115 and crossreactive T cells20,23,26,30,44,78. So far, Wilms’ tumor antigen (WT1), a zinc fi nger
transcription factor overexpressed in many malignancies79, is the only antigen that has repeatedly elicited high-avidity WT1-specifi c T cell responses across HLA-barriers78,80. Possibly, the WT1 T cell epitope (pWT126) induces a unique confi rmation when bound by HLA-A2 molecules, limiting the stimulation of crossreactive T cells. When we stimulated alloHLA-A2-reactive T cells generated using complex-coated DCs with EBV-LCLs naturally expressing HLA-A2, we observed enhanced alloHLA-A2 reactivity as defi ned by increased cytotoxicity and crossreactivity. Re-tuning of T cell avidity in response to the EBV-LCLs’
A. High determinant density model: alloHLA-recognition is mediated by a small subset of donor T cells recognizing alloHLA molecules independent of bound peptide.
B. Multiple binary complex model: alloHLA-recognition is mediated by a large number of different donor T cells recognizing alloHLA/peptide complexes in a peptide-specifi c manner comparable to self-HLA/
peptide-specifi c T cells.
C. Increased degeneracy model: alloHLA-recognition is mediated (in part) by crossreactivity of self-HLA/
peptide-specifi c T cells recognizing alloHLA/peptide complexes with increased degeneracy.
FIGURE 1. THE THREE MODELS OF ALLOHLA-RECOGNITION
116
decreased ligand density as compared to the complex-coated DCs could explain these fi ndings, illustrating the relation between avidity and crossreactivity outlined in paragraph 2.1.2. This model positions crossreactivity as the driving force behind alloHLA-recognition and has considerable consequences for the generation of antigen-specific alloHLA- restricted T cells for adoptive immunotherapy.
5. THE FUTURE OF IMMUNOTHERAPY AFTER HLA-MISMATCHED SCT
5.1 IMMUNOTHERAPY WITH ALLOHLA-RESTRICTED ANTIGEN-SPECIFIC T CELLS
The studies presented in this thesis provide evidence that alloHLA-recognition by T cells is inherently crossreactive. Such crossreactivity seriously hampers the feasibility of generating T cells specific for leukemia- or hematopoietic system-restricted antigens across HLA barriers for the purpose of adoptive cellular immunotherapy.
The currently available technologies can not readily distinguish between crossreactive or antigen-specific alloHLA-restricted T cells, because both phenotypes may closely resemble each other. Stimulating or selecting T cells with single trimeric or tetrameric alloHLA/peptide complexes does not ensure antigen specificity of T cells, because crossreactivity exists at a clonal level (chapter 5). In consequence, truly antigen-specific alloHLA-restricted T cells can only be generated by limiting dilution and extensive in vitro testing. This is a time-consuming and laborious procedure. Moreover, due to the impact of environmental influences on the alloHLA-reactive T cell repertoire, chances of success are difficult to predict and may even vary in time for the same donor. Because many cell divisions are required to obtain sufficient numbers of clonal T cells for immunotherapeutical applications, T cell activity and long-term persistence after infusion into the patient are likely to be impaired81. Therefore, we conclude that the currently available strategies of generating alloHLA-restricted leukemia- or hematopoietic system-restricted antigen-specific T cells are at this stage not applicable in a clinical setting.
5.2 ALTERNATIVE STRATEGIES FOR ADOPTIVE IMMUNOTHERAPY AFTER HLA-MISMATCHED SCT
SCT across HLA barriers can be a clinical necessity for the treatment of a hematological malignancy if a suitable HLA-matched stem cell donor is unavailable. The patient is then at increased risk of relapse, due to obligatory T cell depletion of the stem cell graft and concurring absence of the GvL effect. The clinical question thus remains what therapeutic options such patients can be offered in the case of relapse, given the results presented above.
Several alternative strategies to facilitate adoptive immunotherapy after HLA-mismatched SCT have been proposed over the past years. Unfortunately, our findings affect the feasibility of many of these strategies. For instance, depletion of undesired alloHLA-reactive
117 T cell subsets from T cell populations was shown to retain GvL and anti-viral immunity,
while reducing GvHD incidence in a murine model22,82,83. However, environmental triggers such as microbial infections and exposure to non-infectious antigens that shape the human alloHLA-reactive T cell repertoire are largely absent in laboratory animals routinely kept under very clean conditions. Furthermore, reduction of alloHLA-reactivity following the depletion of alloHLA-reactive T cell subsets may only be temporary due to re-tuning of T cell avidity and preferential expansion of crossreactive T cells upon in vitro culture.
We earlier detected the return of strong alloHLA-responses following alloHLA-reactive T cell depletion upon restimulation with target cells expressing alloHLA26. Thus, it may not be feasible to effectively separate GvHD-inducing T cells from GvL-mediating T cells.
The same mechanisms may hamper strategies for the generation of leukemia-restricted T cells using HLA-mismatched leukemic APCs as stimulators84, or the use of antagonist APLs to inhibit alloHLA-reactivity85.
Two strategies that seem more promising in the long run are TCR gene transfer and the use of umbilical cord blood (UCB).
5.2.1 TCR GENE TRANSFER
Molecular transfer of donor T cell specifi city via the TCR permits the instantaneous generation of a defi ned T cell immunity not present in the autologous T cell repertoire (reviewed86,87). Donor T cell specifi city may thus be redirected to generate hematopoietic system-specifi c or leukemia-associated antigen-specifi c T cells for adoptive immunotherapy after allogeneic SCT. In consequence, a few well characterized TCRs could be used as generic reagents to treat a relatively large number of patients. This is of particular use in the HLA-mismatched setting. TCRs specifi c for the HLA-A2-restricted mHags HA-1 and HA-2 have been successfully introduced into T cells of both HLA-A2pos and HLA- A2neg individuals, and were shown to exert HA-1- and HA-2-specifi c responses without appearance of alloHLA-A2 reactivity in vitro88,89.
The use of TCR gene transfer raises several safety concerns in the setting of adoptive immunotherapy. First, pairing of introduced and endogenous TCRα and TCRβ chains may lead to the formation of new TCRs with unpredictable specifi cities. Second, the effects of TCR transfer may differ depending on the nature of the recipient donor T cells. TCR transfer into naive or suppressor T cells may lead to suppression of the desired T cell response, while end-stage differentiated effector T cells will display limited proliferative and cytotoxic capacities. Third, potentially self-reactive T cells may be activated through the exogenous TCR and thereby induce “bystander” autoimmunity. It was recently proposed to use cytomegalovirus-specifi c T cells as recipients to counter some of these problems. HLA-B7/CMV-specifi c T cells transduced with an HLA-A2/HA-2-specifi c TCR displayed dual specifi city for CMV and for HA-2, effi ciently lysed HLA-A2pos leukemic cells, and showed no signs of broad alloHLA-A2 reactivity90.
As discussed previously, a T cell’s virus specifi city by no means excludes alloHLA- reactivity. Clonal TCR-redirected T cells should therefore be extensively tested for
118
alloHLA-reactivity in each donor - patient combination prior to infusion. The aAPC system we developed may assist in expanding redirected T cells and perhaps even sway T cell avidity in vitro by varying ligand-density4. Co-transduction with a suicide gene may help to control clinical alloHLA-reactivity if re-tuning of T cell avidity in vivo results in new and unexpected alloresponses91. TCR gene therapy may thus open new possibilities for the treatment of leukemia relapse in the future. As yet, TCR gene transfer remains a diffi cult and laborious technique. In addition, random insertion of gene vectors into the genome has been linked to oncogenesis92, while vectors with fi xed insertion patterns are not yet available at GMP grade. This and many other technical problems that remain to be solved make introduction of TCR transfer into the clinic at short notice unlikely.
5.2.2 UMBILICAL CORD BLOOD TRANSPLANTATION
A safe and feasible alternative to adult HLA-mismatched SCT involves the use of umbilical cord blood as a stem cell source (reviewed93,94). UCB-SCT is associated with decreased risk and severity of GvHD after SCT with HLA-matched or 1-2 HLA-mismatched grafts as compared to adult SCT. The lower GvHD incidence is thought to result from the relatively naive phenotype and more stringent activation requirements of co-transplanted UCB-derived T cells as compared to adult bone marrow- or peripheral blood-derived T cells95. Yet, relapse rates after UCB-SCT are comparable to those after adult SCT, indicating the presence of GvL activity. HA-1-specifi c T cells of child origin were recently detected in UCB samples from HA-1neg children of HA-1pos mothers96. The generation of tumor- specifi c alloHLA-restricted T cells from UCB could perhaps be associated with decreased induction of alloHLA-reactivity in parallel. Unfortunately, cord blood units are as yet not suffi ciently large to allow both transplantation and adoptive immunotherapy. Ex vivo expansion systems are currently under investigation so that immunotherapy after HLA- mismatched UCB-SCT may be possible in the future.
5.3 CONCLUDING REMARKS
In this thesis we have explored the alloHLA-A2-reactive T cell repertoire and have tested the feasibility of generating alloHLA-A2-restricted T cells specifi c for the hematopoietic system-restricted mHags HA-1 or HA-2 for adoptive immunotherapy. Our results show that the alloHLA-A2-restricted T cell repertoire is inherently crossreactive. This crossreactivity hampers the generation of alloHLA-restricted mHag-specifi c T cells and limits the feasibility of adoptive immunotherapy with T cells after HLA-mismatched SCT in general. Until strategies such as TCR gene transfer and use of UCB have been optimized, we suggest using the shared HLA molecules instead of the alloHLA molecules for the purpose of adoptive immunotherapy in the HLA-mismatched setting. HLA-mismatched donor - patient combinations must share at least one HLA haplotype to enable interactions between donor T cells and host APCs. Therefore, donor T cells can be targeted to hematopoietic system-specifi c mHags or leukemia-associated antigens presented in the
119 context of a shared HLA molecule, analogous to the set-up for immunotherapy after
HLA-matched SCT22,97-99. While these donor T cells will display specifi city for the shared HLA molecules facilitating the generation of high-avidity T cells, crossreactivity with the patient’s alloHLA molecules may still occur. Thus, a fully HLA-matched SCT remains the best setting for cellular adoptive immunotherapy.
REFERENCES
1. Walter S, Herrgen L, Schoor O, et al. Cutting edge: predetermined avidity of human CD8 T cells expanded on calibrated MHC/anti-CD28-coated microspheres. J Immunol. 2003;171:4974-4978.
2. Oelke M, Maus MV, Didiano D, June CH, Mackensen A, Schneck JP. Ex vivo induction and expansion of antigen-specifi c cytotoxic T cells by HLA-Ig-coated artifi cial antigen-presenting cells. Nat Med. 2003;9:619-624.
3. Dutoit V, Rubio-Godoy V, Pittet MJ, et al. Degeneracy of antigen recognition as the molecular basis for the high frequency of naive A2/Melan-a peptide multimer(+) CD8(+) T cells in humans. J Exp Med. 2002;196:207-216.
4. Schilbach K, Kerst G, Walter S, et al. Cytotoxic minor histocompatibility antigen HA-1-specifi c CD8+ effector memory T cells: artifi cial APCs pave the way for clinical application by potent primary in vitro induction. Blood. 2005;106:144-149.
5. Maus MV, Riley JL, Kwok WW, Nepom GT, June CH. HLA tetramer-based artificial antigen-presenting cells for stimulation of CD4+ T cells. Clin Immunol. 2003;106:16-22.
6. Bromley SK, Burack WR, Johnson KG, et al. The immunological synapse. Annu Rev Immunol. 2001;19:375-396.
7. Greenwald RJ, Freeman GJ, Sharpe AH. The B7 family revisited. Annu Rev Immunol. 2005;23:515-548.
8. Curtsinger J, Deeths MJ, Pease P, Mescher MF. Artifi cial cell surface constructs for studying receptor-ligand contributions to lymphocyte activation. J Immunol Methods. 1997;209:47-57.
9. Kim JV, Latouche JB, Riviere I, Sadelain M. The ABCs of artifi cial antigen presentation. Nat Biotechnol. 2004;22:403-410.
10. Shuford WW, Klussman K, Tritchler DD, et al. 4-1BB costimulatory signals preferentially induce CD8+ T cell proliferation and lead to the amplifi cation in vivo of cytotoxic T cell responses. J Exp Med. 1997;186:47-55.
11. Diehl L, van Mierlo GJ, den Boer AT, et al. In vivo triggering through 4-1BB enables Th-independent priming of CTL in the presence of an intact CD28 costimulatory pathway. J Immunol. 2002;168:3755-3762.
12. Laderach D, Movassagh M, Johnson A, Mittler RS, Galy A. 4-1BB co-stimulation enhances human CD8(+) T cell priming by augmenting the proliferation and survival of effector CD8(+) T cells. Int Immunol. 2002;14:1155-1167.
13. Hendriks J, Gravestein LA, Tesselaar K, van Lier RA, Schumacher TN, Borst J. CD27 is required for generation and long-term maintenance of T cell immunity. Nat Immunol. 2000;1:433-440.
14. Zeng R, Spolski R, Finkelstein SE, et al. Synergy of IL-21 and IL-15 in regulating CD8+ T cell expansion and function.
J Exp Med. 2005;201:139-148.
15. Wei ML, Cresswell P. HLA-A2 molecules in an antigen-processing mutant cell contain signal sequence-derived peptides.
Nature. 1992;356:443-446.
16. Bangia N, Cresswell P. Stoichiometric tapasin interactions in the catalysis of major histocompatibility complex class I molecule assembly. Immunology. 2005;114:346-353.
17. Smith PA, Brunmark A, Jackson MR, Potter TA. Peptide-independent recognition by alloreactive cytotoxic T lymphocytes (CTL). J Exp Med. 1997;185:1023-1033.
18. Faber LM, van der Hoeven J, Goulmy E, Luxemburg-Heijs SA, Willemze R, Falkenburg JHF. Recognition of clonogenic leukemic cells, remission bone marrow and HLA-identical donor bone marrow by CD8+ or CD4+ minor histocompatibility antigen-specifi c cytotoxic T lymphocytes. J Clin Invest. 1995;96:877-883.
19. Wang W, Man S, Gulden PH, Hunt DF, Engelhard VH. Class I-restricted alloreactive cytotoxic T lymphocytes recognize a complex array of specifi c MHC-associated peptides. J Immunol. 1998;160:1091-1097.
20. Sadovnikova E, Parovichnikova EN, Savchenko VG, Zabotina T, Stauss HJ. The CD68 protein as a potential target for leukaemia-reactive CTL. Leukemia. 2002;16:2019-2026.
21. Munz C, Obst R, Osen W, Stevanovic S, Rammensee HG. Alloreactivity as a source of high avidity peptide-specifi c human CTL. J Immunol. 1999;162:25-34.
22. Amrolia PJ, Muccioli-Casadei G, Yvon E, et al. Selective depletion of donor alloreactive T cells without loss of antiviral or antileukemic responses. Blood. 2003;102:2292-2299.
23. Amrolia PJ, Reid SD, Gao L, et al. Allorestricted cytotoxic T cells specifi c for human CD45 show potent antileukemic activity. Blood. 2003;101:1007-1014.
24. Altman JD, Moss PA, Goulder PJ, et al. Phenotypic analysis of antigen-specifi c T lymphocytes. Science. 1996;274:94-96.
25. Ogg GS, Jin X, Bonhoeffer S, et al. Quantitation of HIV-1-specifi c cytotoxic T lymphocytes and plasma load of viral RNA. Science. 1998;279:2103-2106.
26. Mutis T, Blokland E, Kester M, Schrama E, Goulmy E. Generation of minor histocompatibility antigen HA-1-specifi c cytotoxic T cells restricted by nonself HLA molecules: a potential strategy to treat relapsed leukemia after HLA- mismatched stem cell transplantation. Blood. 2002;100:547-552.
120
27. Mutis T, Ghoreschi K, Schrama E, et al. Effi cient induction of minor histocompatibility antigen HA-1-specifi c cytotoxic T cells using dendritic cells retrovirally transduced with HA-1-coding cDNA. Biol Blood Marrow Transplant. 2002;8:412-419.
28. Choudhuri K, Wiseman D, Brown MH, Gould K, van der Merwe PA. T cell receptor triggering is critically dependent on the dimensions of its peptide-MHC ligand. Nature. 2005;436:578-582.
29. Kissler S, Anderton SM, Wraith DC. Cross reactivity and T cell receptor antagonism of myelin basic protein-reactive T cells is modulated by the activation state of the antigen-presenting cell. J Autoimmun. 2002;19:183-193.
30. Moris A, Teichgraber V, Gauthier L, Buhring HJ, Rammensee HG. Cutting edge: characterization of allorestricted and peptide-selective alloreactive T cells using HLA-tetramer selection. J Immunol. 2001;166:4818-4821.
31. Akatsuka Y, Nishida T, Kondo E, et al. Identifi cation of a polymorphic gene, BCL2A1, encoding two novel hematopoietic lineage-specifi c minor histocompatibility antigens. J Exp Med. 2003;197:1489-1500.
32. Verdijk RM, Kloosterman A, Pool J, et al. Pregnancy induces minor histocompatibility antigen-specifi c cytotoxic T cells:
implications for stem cell transplantation and immunotherapy. Blood. 2004;103:1961-1964.
33. Tekolf WA, Shaw S. In vitro generation of cytotoxic cells specifi c for human minor histocompatibility antigens by lymphocytes from a normal donor potentially primed during pregnancy. J Exp Med. 1983;157:2172-2177.
34. van Rood JJ, Eernisse JG, van Leeuwen A. Leucocyte antibodies in sera from pregnant women. Nature. 1958;181:1735-1736.
35. Aluvihare VR, Kallikourdis M, Betz AG. Tolerance, suppression and the fetal allograft. J Mol Med. 2005;83:88-96.
36. Tilburgs T, Roelen DL, van der Mast BJ, et al. Differential distribution of CD4(+)CD25(bright) and CD8(+)CD28(-) T cells in decidua and maternal blood during human pregnancy. Placenta. 2006;27 Suppl A:S47-S53.
37. Levings MK, Roncarolo MG. T-regulatory 1 cells: a novel subset of CD4 T cells with immunoregulatory properties. J Allergy Clin Immunol. 2000;106:S109-S112.
38. Pittet MJ, Gati A, Le Gal FA, et al. Ex vivo characterization of allo-MHC-restricted T cells specifi c for a single MHC- peptide complex. J Immunol. 2006;176:2330-2336.
39. Ernst B, Lee DS, Chang JM, Sprent J, Surh CD. The peptide ligands mediating positive selection in the thymus control T cell survival and homeostatic proliferation in the periphery. Immunity. 1999;11:173-181.
40. Goldrath AW, Bevan MJ. Low-affi nity ligands for the TCR drive proliferation of mature CD8+ T cells in lymphopenic hosts. Immunity. 1999;11:183-190.
41. Mason D. A very high level of crossreactivity is an essential feature of the T cell receptor. Immunol Today. 1998;19:395-404.
42. Wucherpfennig KW. T cell receptor crossreactivity as a general property of T cell recognition. Mol Immunol.
2004;40:1009-1017.
43. Wilson DB, Wilson DH, Schroder K, et al. Specifi city and degeneracy of T cells. Mol Immunol. 2004;40:1047-1055.
44. Dutoit V, Guillaume P, Romero P, Cerottini JC, Valmori D. Functional analysis of HLA-A*0201/Melan-A peptide multimer+ CD8+ T cells isolated from an HLA-A*. Cancer Immun. 2002;2:7.
45. Rudolph MG, Stanfi eld RL, Wilson IA. How TCRs bind MHCs, peptides, and coreceptors. Annu Rev Immunol.
2006;24:419-466.
46. Reiser JB, Gregoire C, Darnault C, et al. A T cell receptor CDR3beta loop undergoes conformational changes of unprecedented magnitude upon binding to a peptide/MHC class I complex. Immunity. 2002;16:345-354.
47. Ding YH, Baker BM, Garboczi DN, Biddison WE, Wiley DC. Four A6-TCR/peptide/HLA-A2 structures that generate very different T cell signals are nearly identical. Immunity. 1999;11:45-56.
48. Wu LC, Tuot DS, Lyons DS, Garcia KC, Davis MM. Two-step binding mechanism for T cell receptor recognition of peptide MHC. Nature. 2002;418:552-556.
49. Maverakis E, van den EP, Sercarz EE. Self-reactive T cells and degeneracy of T cell recognition: evolving concepts-from sequence homology to shape mimicry and TCR fl exibility. J Autoimmun. 2001;16:201-209.
50. Hiemstra HS, Schloot NC, van Veelen PA, et al. Cytomegalovirus in autoimmunity: T cell crossreactivity to viral antigen and autoantigen glutamic acid decarboxylase. Proc Natl Acad Sci U S A. 2001;98:3988-3991.
51. Martin R, Gran B, Zhao Y, et al. Molecular mimicry and antigen-specifi c T cell responses in multiple sclerosis and chronic CNS Lyme disease. J Autoimmun. 2001;16:187-192.
52. Gautam AM, Pearson CI, Smilek DE, Steinman L, McDevitt HO. A polyalanine peptide with only fi ve native myelin basic protein residues induces autoimmune encephalomyelitis. J Exp Med. 1992;176:605-609.
53. Bhardwaj V, Kumar V, Geysen HM, Sercarz EE. Degenerate recognition of a dissimilar antigenic peptide by myelin basic protein-reactive T cells. Implications for thymic education and autoimmunity. J Immunol. 1993;151:5000-5010.
54. Li Y, Huang Y, Lue J, Quandt JA, Martin R, Mariuzza RA. Structure of a human autoimmune TCR bound to a myelin basic protein self-peptide and a multiple sclerosis-associated MHC class II molecule. EMBO J. 2005;24:2968-2979.
55. Davis MM, Boniface JJ, Reich Z, et al. Ligand recognition by alpha beta T cell receptors. Annu Rev Immunol.
1998;16:523-544.
56. Gagnon SJ, Turner RV, Shiue MG, Damirjian M, Biddison WE. Extensive T cell receptor cross reactivity on structurally diverse haptenated peptides presented by HLA-A2. Mol Immunol. 2006;43:346-356.
57. Hahn M, Nicholson MJ, Pyrdol J, Wucherpfennig KW. Unconventional topology of self-peptide-major histocompatibility complex binding by a human autoimmune T cell receptor. Nat Immunol. 2005;6:490-496.
58. Sloan-Lancaster J, Allen PM. Altered peptide ligand-induced partial T cell activation: molecular mechanisms and role in T cell biology. Annu Rev Immunol. 1996;14:1-27.
59. Huseby ES, White J, Crawford F, et al. How the T cell repertoire becomes Peptide and MHC specifi c. Cell. 2005;122:247-260.
60. Tanchot C, Barber DL, Chiodetti L, Schwartz RH. Adaptive tolerance of CD4+ T cells in vivo: multiple thresholds in response to a constant level of antigen presentation. J Immunol. 2001;167:2030-2039.
61. Anderton SM, Wraith DC. Selection and fi ne-tuning of the autoimmune T cell repertoire. Nat Rev Immunol. 2002;2:
487-498.
