On cross-priming of MHC class I-specific CTL: rule or exception?

Minor histocompatibility antigens Nucleoprotein Rat insulin promotor Transgenic Glycoprotein It has been generally accepted as a rule that peptides binding to MHC class I molecules are derived from proteins synthesized within the cell, including peptides of genuine cellular origin as well as those encoded by viruses or other intracellular infectious agents. In contrast, MHC class II molecules, in general, bind peptides derived from exogenous proteins that have been processed viaphago-lysozomes and replace the invariant chain clip fragment associated with class II molecules 1, 2. MHC class I restriction correlates with CD8+ cytotoxic T cells (CTL) and class II restriction with CD4+ T (Th) cells. However, soon after the initial formulation of these general rules, differential MHC restriction as well as the divisionof labor between endogenous versus exogenous loading of MHC class I and class II molecules were questioned 3–13. The subject of this editorial is so-called "cross-presentation" and "cross-priming" as well as "cross-tolerance", which may be defined as the presentation of exogenous antigen via MHC class I for generating stimulatory or tolerogenic responses in CD8+ T cells, respectively (Table 1) 10; accordingly, exogenous antigen has "crossed" over to the endogenous pathway to gain access to MHC class I. The use of the term "cross-priming" for peptides presented on MHC class II molecules is not warranted because this is the conventional pathway for class II presentation (e.g. 14). Recently, MHC class I-specific presentation of exogenous antigen to induce cross-priming, cross-presentation, or cross-tolerance has become viewed by many as a general and important phenomenon 5, 9, 10, 13, 15. This emerging view is based on the concept that activation of CD4+ Th cells or CD8+ T CTL requires antigen (signal 1) plus costimulation (signal 2), whereas encounters with signal 1 alone result in anergy or deletion 16–19. This wasthe basis for the proposal: Unless cross-presentation plays an essential role, self-encoded antigens, which are extralymphatic, could not render T cells tolerant, and viruses, which do not infect professional antigen-presenting cells (APC, expressing both antigen as signal 1 and signal 2) or tumor cells (that do not express signal 2) could not elicit MHC class I-restricted CTL 5, 9, 10, 13. In biology, particularly in immunology, nothing is impossible. However, as a result of co-evolution of pathogens and their hosts, there has been selection of frequent or likely pathways and effector mechanisms against unlikely rare forms and exceptions. Therefore, the key question here is: Is cross-priming via MHC class I a general and essential characteristic of physiologically important CD8+ CTL cytolytic or tolerogenic responses or, alternatively, is this process an exception that, perhaps under special conditions, can be exploited therapeutically to induce responses that otherwise would be difficult to generate (Table 1)? The first evidence for this concept was obtained in early experiments by Bevan 5, 6 and subsequently by Simpson and Gordon 20 and von Boehmer (unpublished observations) using so-called minor histocompatibility antigens (miH) as targets for cytotoxic T cells. miH represent allelic differences of usually numerous host protein-derived peptides that are presented by MHC class I (also by MHC class II, but those are rarely studied 21, 22). In some cases, miH differences may include antigens encodedby endogenous retroviruses 23, 24. The original experiments and postulates were not based upon 2-signal theories or the exclusive inductive capacity of so-calledprofessional APC, but were based on empirical findings involving skin graft rejection or CTL responses. For example, if an H-2b BALB.B mouse was immunized with H-2b C57BL/6 (B6) splenocytes, a minor histocompatibility-specific CTL response was measurable on Con A- or LPS-treated lymphoblast cells derived from the donor B6 H-2b. If the same experiment was performed by immunization of (H-2d BALB/c×H-2b BALB.B)F1 recipients with H-2b B6 spleen cells, followed by restimulation in vitro with either H-2b B6 or H-2d B10.D2 lymphocytes from H-2 congenic mouse strains, CTL activity specifically restricted to the H-2d haplotype was also observed. This "cross-primed" CTL activity of the F1-host was detectable in vitro using LPS or Con A blasts as target cells, but apparently was not readily found against non-mitogen-treated lymphocytes or fibroblast targets. The role of Th in skin graft rejection or CTL responses in vitro was subsequently studied in detail and positively demonstrated 21, 25. Interestingly, even in these very early experiments, the use of tumor cells instead of spleen cells revealed no such cross-priming under similar experimental conditions 5, 6. This finding was interpreted as reflecting a "narrow" spectrum of specific responses whereby the injected tumor cells had expanded to large numbers and perhaps selected one dominant CTL subspecificity 26–29. Subsequent experiments used latex beads coated with a protein antigen 30, 31, protein antigens forced into the cytosol of cells via hypotonic shock 32, or, more recently, immune complexes 15, 33 (or exosomes 34). These approaches avoided the uncertainties of undefined miH, for which neither the precise epitope nor the relative distribution and expression was known or controllable (even in so-called miH-congenic mouse strains). These more recent experiments formally showed that antigens entering cells from the outside, under special experimental conditions, could enter the MHC class I presentation pathway. Thus, cross-presentation and cross-priming are possible; but are they phenomena of general importance or an exception? Studies with melanoma or tumor virus antigen-associated CTL responses are often quoted as the basis for the in vivo relevance of cross-presentation and cross-priming. The experiments performed in Pardoll's laboratory used melanoma B16 mutant cells that did or did not express MHC class I antigens to immunize syngeneic H-2b mice 35. Specific CTL responses showing about 25–30% cytolysis were observed in vitro at high effector to target cell ratios (100:1), with 10% of control targets being lysed. Protection against tumor take was alsodocumented in vivo, but the studies could not attribute the effect exclusively to CTL effector mechanisms. In fact, later studies suggested that alternative effector mechanisms were also important 36. In another study, colon carcinoma H-2b tumors expressing the influenza nucleoprotein (NP) were used to immunize F1 (H-2b × H-2d) → H-2b or H-2d chimeras; chimeras were then tested for cross-primed CTL after restimulation in vivo. Peptide-specific lysis assays (measuring the difference between lysis ofpeptide-loaded versus unloaded target cells) tested at a 100:1 ratio yielded 10–20% specific lysis by cross-primed CTL 35. The experiments by Melief and colleagues 37, 38 on adenovirus 5E1A or B + Db (Ad5E1A or B) specific responses are based on AdE1A or B adenovirus-transformed tumor cells of BALB/c (H-2d) origin. C57BL/6 (H-2b) mice immunized twice with these tumor cells were restimulated in vitro with irradiated E1A or B + H-2b tumor cells for 5–6 days. They were then tested against Europium-labeled targets in a microassay. AdE1A or B expressing targets were lysed about ten times better than control targets, indicating that allogeneic AdE1A+ or B+ tumor cells had induced a CTL response.In vivo protection against tumor take was tested by similar vaccination protocols. By 14 days after immunization, up to 75% of the animals were protected against a challenge with 5×105 Ad5E1+ ras transformed tumor cells. The role of mechanisms other than CTL killing, for example protective Th responses, was initially not determined, but addressed in later studies 39. While the studies demonstrated that alloresponses as such are (alone) not responsible forthe peptide specificity seen, they discussed the inherent problems of a widely cross-reactive alloresponse and formulated carefully "that cross-priming is most likely sufficient". Nevertheless, they conclude "that completely allogeneic tumor cells can be successfully used for the induction of protective anti-tumor immunity and hold promise for the implementation of allogeneic tumor cell basedvaccines". More recently, Heath and collaborators 10, 40–42 used transgenic mice expressing ovalbumin (OVA) under the rat-insulin promotor (RIP-OVA) to show that transgenic (tg) OVA-Db-specific T cell receptor (TCR) expressing CD8+ T lymphocytes were activated in draining lymph nodes of OVA-overexpressing pancreatic islets. Thiswas interpreted as cross-presentation of a peripheral "self" antigen in the draining lymph nodes based on the assumption that only APC could pick up apoptotic or necrotic islet cells, cross-processantigen, and induce tgTCR+ CD8 T cells in the draining lymph node. A potential role of Th was not considered in the original studies, but was subsequently confirmed 43, 44. The response was also dependent on antigen dose and on very high numbers of TCR-tg CTL to achieve the described phenotype 10. These experiments caused great enthusiasm for cross-presentation/cross-priming/cross-tolerance as a general and important mechanism 1 The fact that OVA is the most favored protein used for model experiments on cross-priming may have a specific reason. Perhaps it is linked to a very special property of OVA, which is, after all, a transport protein capable of binding to many other proteins, but this remains to be carefully analyzed. Shortly following this report, two independent experiments were published, with many positive commentaries, which demonstrated in vitro that viral antigens (seemingly independent of an infectious process) could be taken up by APC and cross-presented via class I in the cases of influenza viruses 45 and polio virus 46. Similar evidence had come earlier from experiments using hepatitis B surface (HBs) (but not HB core) particles 47 or undefined preparations of recombinant viral protein mixed with debris 48 (but, interestingly, not without debris) from a bacculovirus expression system 49. In contrast to the earlier miH model systems, the later experiments during the 1990s usinglatex beads, hypotonic shock or viral particles defined CTL responses at the peptide level. Interestingly, and surprisingly, none of these later investigations actually demonstrated cross-priming using classical F1 experiments to show cross-presentation via F1 APC. Experiments not showing cross-priming or studies that falsify predictions of the general concept. Experimental evidence has accumulated, also on the negative side. Early H-Y-specific CTL experiments only occasionally showed cross-priming as predicted, and often the CTL activities obtained were low 20. This was interpreted as an example of class I immune response (Ir) regulation implying additional limiting roles of Th cells or that no correct presentable peptides were generated 21. Similar evidence came from a careful study of H43a versus H43b miH-congenic mice, which found specific CTL responses without cross-priming 50. A third example was provided by Mx-congenic mouse strains, where no cross-priming could be demonstrated 51. More recently, transformed fibroblast cells L929 (H-2k) were transfected with the NP of LCMV and tested with or without additional transfection with H-2Ld capable of presenting LCMV-NP118–126 52 (Table 2). When BALB/c (H-2d) mice were immunized with a wide variety of protocols using the allogeneic transfected cells, no Ld+NP118–126-specific CTL were generated except when 106 living L929 (H-2k) NP118–126+Ld-positive fibroblasts were used. This indicated that CTL were induced directly by these fibroblasts, and not via cross-priming. Comparable results were obtained with tumor cells of epithelial, mesenchymal or melanoma origin from H-2b B6 mice (Table 2). These tumors expressed tumor antigens plus the transfected glycoprotein (GP) of LCMV as an artificial tumor antigen. The tumor cells promptly induced Db-GP33–41 CTL in B6 mice when 104 living cell were injected into the spleen; this efficiency is comparable to APC transgenic for the same GPLCMV. In contrast, even after multiple injections into F1 mice of apoptotic, necrotic or living cells, the tumor cells all failed to reveal cross-priming. For each tumor example, the readout was specific for defined class I-presented peptides. As pointed out above, the lack of cross-priming with tumor cells might have been expected from earlier studies 6. Therefore, the same experiments were repeated using spleen cells from H-2b or H-2d transgenic mice expressing the LCMV GP using an ubiquitous promoter. Again, no significant cross-priming was observed under conditions where direct priming was very efficient 53. Taken together, these findings suggested two general conclusions (Table 1): (1) Whenever large quantities 15, 30, 31, 54 of particulate-lipid-containing materials 45–49, 55 are used, which allow antigen to gain access to the cytosol, "cross-priming" can be shown; but formal F1 experiments appear to be difficult or unfeasible. (2) Whenever a multitude of miH are used, cross-priming is readily shown with "physiological materials", whereas ifonly a single miH or a defined viral antigen, exhibiting only very rare and few CTL epitopes, are used, cross-priming is not seen in a classical F1 experiment. How can these differing observations be explained? Do they reflect methodological/experimental differences? Or do they suggest separate pathways of efficient antigen processing that depend on the various antigens used? Cross-presentation via MHC class I? 1. Two signal hypothesis demands obligatory "professional" (i.e. signal 2 positive, upregulated, "mature") antigen-presenting cells (APC) for T cell induction to become effector T cells. 2. Negative selection is mandatory for all T cells specific for self-encoded peptides. 3. T cell induction to become effector T cells and T cell deletion-anergy-tolerization are two distinct processes; absence of signal 2 on APC is a negative signal (i.e. immune regulation by suppression). An alternative view 1. Any antigen-expressing cell reaching secondary lymphatic organs (or equivalents for some, such as intraepithelial micropatches, etc.) induces effector T cells. 2. No negative selection (equivalent to exhaustive induction) necessary for T cells specific for self or non-self encoded peptides that stay extralymphatic (ignorance). 3. T cells are inducible only: induction of <100% of precursor T cells to become effector T cells causes elimination or control down to very low levels of antigen ("immunity" = maintenance of activated T cells by low level persistence or re-encountered antigen). Induction of 100% of precursor T cells deletes/exhaustively induces all precursor T cells due to the 3–4 d half-life of effector T cells. Questions and unresolved problems of experiments on cross-priming 1. Why are necessary conditions for cross-priming special, i.e.: great quantities of proteins, or lipids, or lipoproteins, or aggregates, and/or corpuscular materials (local concentrations high, mostly non-physiological in the sense that conditions cannot be easily reached by apoptotic or necrotic cells in vivo). 2. Inactivation of virus materials often poorly controlled in vitro and in vivo. 3. Very sensitive cytotoxicity read out (lymphoblast targets, high spontaneous 51Cr released, shallow slopes of Cr-release, high E:T ratios), or T cell proliferation. 4. Multiple (very many miH or other) targets, often undefined at peptide level. 5. Protective capacity of effector T cells often not tested in vivo, or exclusive involvement of CTL unclear. 6. Whenever a single defined cellular antigen with defined T epitope peptides is cross-priming is to or not in the classical F1 experiment. of T class II MHC is often T can be limiting for CTL responses. of or T show responses via of mice or or with cells transfected CTL response Experiments in C57BL/6 (H-2b) CTL 1 104 + + 2 104 fibroblasts transfected with LCMV GP classical + + LCMV fibroblasts or injected or over Experiments in CTL + completely allogeneic fibroblasts (H-2k) protocols as for 106 allogeneic (H-2k) fibroblasts transfected with plus + + the original miH In most cases, of miH differences were in the responses. In these experiments all used target cells to induced CTL responses. The of such blasts are that the MHC is that cellular antigens including potential endogenous viruses are also expressed at high and that spontaneous 51Cr is high (i.e. for In on target cell lysis presented for and targets, as the difference between these blasts to or fibroblasts or other cells, the difference in MHC antigen expression levels is The problems of target are by the usually shallow slopes of 51Cr from lymphoblast targets. By or targets reveal an and directly of and target cell as pointed out above, when used a single defined miH, including antigens, cross-presentation and cross-priming cannot be easily for these might be that the numerous of CTL with the high of the target cell may reveal of CTL that are in but are of no in vivo. This may for cross-priming has not formally been with CTL responses in classical F1 In this it is to that cross-reactive T cell have been with for miH and for or miH and but have not been observed for protective T cells generated in vivo antigen-expressing fibroblasts or from these is that the used to so-called cross-primed CTL is low in if not all, When such are tested for in protection in vivo conditions where only CD8+ CTL and neither T cells nor are cross-priming has not been under conditions of extralymphatic viral that this not the that with antigen and at of antigen, cross-priming can be forced The role of Th in generating CTL and cross-priming has been discussed experiments suggest that there is a more for Th in cross-priming than in direct priming because antigen are Experiments using 21, adenovirus or virus proteins and miH 38 more also for OVA an important role for Th cells. In a for Th cells for CTL induction has also been for direct priming under conditions where antigen and may be For example, of mice with viruses that induces CTL responses in the absence of Th cells. In other conditions where viruses do not or only in mice after influenza virus or virus T is an important for CTL induction Therefore, the question in experimental conditions, limiting of T via the conventional exogenous MHC class II were instead of cross-presentation and cross-priming via MHC class I. This has been discussed in general, has not been For example, in the model Th cells appear to be in an important Therefore, some of the findings may actually reflect activation of the large numbers of precursor transgenic CTL than MHC class I cross-presentation and cross-priming by APC in the draining lymph node. not an the demonstrated in these experiments were dependent on both high OVA antigen expression in the their as well as a high of transgenic of these conditions activation a role of Th in limiting CTL responses may the restimulation of CTL responses as well as the cross-primed responses in melanoma or model CTL induction against defined peptides have been for influenza NP preparations of viral antigens expressed in bacculovirus influenza virus 13, 45 and 46. The of particles necessary cell to induce such cross-presentation and cross-priming in vivo or in vitro is but is on the of 55 example, 1 about 106 and about of these experiments cannot virus with cell These and other experiments using hypotonic shock or latex beads show that it is to antigens into the pathways and MHC class I. However, independent experiments suggest that in vivo tumor cells or virus under conditions apoptotic and necrotic or antigen-expressing cells from transgenic mice to the phenomenon of cross-presentation and cross-priming, particularly when tested in the classical F1 experiment. all one may conclude that, it is to cross-presentation and cross-priming under selected special conditions, this is not a phenomenon that conditions in vivo. Therefore, cross-presentation and cross-priming may perhaps be for of conditions, this evidence and be to that cross-presentation and cross-priming via MHC class I are essential and of CTL and It is that there has been of may be cases of Recently, there has been an not only of signal but also about T cells, negative T cell and as if they had not (or even of during the In fact, "cross-priming" the on The fact that it was to induce that even in rare cases may control B cell tumors the concept of the control of the which there no experimental The difference between is under experimental versus is a The that tumor cell expressing melanoma antigens can CTL in of living tumor cells directly presenting MHC class is not in the evidence that such is unlikely to be in vivo because it is While this can be in an with the completely or apoptotic or necrotic tumor materials to generate an exclusively cross-primed effector T cell where the induction of T or is important in tumors or viruses, protection by other mechanisms than CTL and not via MHC class I may be It is also important to that exogenous antigens into the MHC class I pathway by special experimental may cross-presentation via MHC class I. However, this is not a process that likely for both conventional of induction of CTL against viruses those not or against peripheral or mesenchymal evidence that cross-presentation is unlikely to an important role in of MHC class I-restricted CTL or in deletion or against self antigens expressed exclusively