T cell epitope–specific immune therapy for rheumatic diseases

Introduction

Rheumatoid inflammation is a chronic and self-amplifying autoimmune process that ultimately leads to destruction of cartilage and bone (1). The pathophysiologic development of inflammation is complex and is the outcome of overlapping, reverberating processes that involve both the adaptive and the innate arms of the immune system (2–7). The events triggering these autoimmune mechanisms are thought to be the result of a combination of genetic predisposition and certain environmental factors (8, 9).

The dramatic progress gained in molecular immunology has enabled the evolution from traditional pharmacologic strategies of aggressive immune suppression to biologic-based therapies aimed at addressing the pathophysiologic process more directly. So far, the greatest progress has been achieved in the area of controlling individual cytokine pathways that contribute to rheumatoid inflammation. In particular, biologic agents aimed at interfering with tumor necrosis factor α (TNFα), interleukin-1 (IL-1), and, more recently, IL-6 have been very successful in various clinical settings (10, 11). None of those strategies, however, induces a sustained remission, and therefore none can fully restore homeostasis to the immune system. Consequently, life-long treatment with such agents is necessary, which entails considerable costs and increases the risk of long-term side effects (12).

Rising awareness of these problems has created a consensus on the need for a therapeutic strategy that will capitalize more comprehensively on our understanding of the immunopathology of rheumatic diseases. In particular, pathways leading to adaptive autoimmunity need to be better exploited. Ideally, such approaches should be crafted so that they are complementary to the current cytokine-directed therapies. We will discuss herein the current status of therapeutic strategies aimed at correcting T cell–mediated inflammation. In particular, we will discuss our perspective regarding a T cell epitope–specific approach to restoration of naturally occurring mechanisms that modulate inflammation.

Insufficiency of T cell blockade alone

Previous studies have shown that a strategy of simply depleting T cells does not offer a solution, because depleting antibody therapy against CD4 has not been successful in controlling inflammation in rheumatoid arthritis (RA) (13, 14) (Table 1). Coating of the CD4 molecule, rather than depletion of CD4 cells, seems to have more beneficial effects (15). An alternative option is to modulate T cell activation by blockade of T cell costimulation.

Table 1. Experience with T cell–directed intervention in rheumatic diseases*

Compound	Mode of action	Effect	Refs.
* CTLA-4Ig = cytotoxic T lymphocyte–associated antigen 4–IgG1 fusion protein; MTX = methotrexate; HSPs = heat-shock proteins; RA = rheumatoid arthritis; gp39 = glycoprotein 39.
Chimeric anti-CD4	CD4 cell depletion	No clinical efficacy	13, 14, 70
Anti-CD4 monoclonal antibody (Keliximab)	CD4 coating	Limited efficacy	15
CTLA-4Ig (abatacept)	Inhibition of T cells	Efficacious with MTX	17
Escherichia coli lysate (OM-89)	Oral tolerance to HSPs?	Limited efficacy in RA	71–75
Type II collagen	Tolerance to collagen	Limited efficacy	24, 76, 77
HLA–DR4–gp39	Tolerance to gp39	Safe, no clinical side effects known	21
dnaJP1	Mucosal tolerization of Treg	Safe, induces immune tolerance, in phase II trial	45

T cells need at least 2 signals in order to become fully activated. Signal 1 requires engagement of the T cell receptor (TCR) on T cells within a complex of peptide and major histocompatibility complex (MHC) on the antigen-presenting cell, whereas signal 2 involves the binding of a costimulatory molecule on a T cell with its ligand on the antigen-presenting cell. Cytotoxic T lymphocyte–associated antigen 4 (CTLA-4) is expressed on activated T cells and binds with high affinity to CD80 and CD86 on antigen-presenting cells, and thus blocks the engagement of these molecules with CD28 (16). This abrogates the optimal delivery of signal 2 and thus contributes to a natural down-regulation of the activated T cell. In addition, CTLA-4 can directly inhibit T cell activation by reducing IL-2 production and IL-2 receptor expression and arresting T cells in the G1 phase of the cell cycle (16).

A fusion protein of the external domain of CTLA-4 containing the heavy chain of human IgG1 (CTLA-4Ig) was constructed. CTLA-4Ig binds to CD80 and CD86 on antigen-presenting cells, thus mimicking the natural down-regulation of T cell activation. Preclinical studies using various animal models of autoimmune diseases have demonstrated the efficacy of CTLA-4Ig. A placebo-controlled trial in RA patients receiving methotrexate showed improvement of disease activity in patients treated with CTLA-4Ig (17). This result indicates that T cells apparently still play a prominent role at later stages of the inflammatory process in RA.

Obviously, since this type of treatment is aimed at blocking T cell activation in general, it lacks specificity. In fact, both T cells that play a role in synovial inflammation and T cells that are recruited to mount a physiologic immune response are targeted. This means that the duration and the dose of the treatment have to be carefully balanced against the risk of infection, cancer, and other unwanted side effects. It would certainly be preferable to control only those lymphocytes that are responsible for the development of synovial inflammation.

Experience with trigger antigen–specific immune therapy

Specifically targeting T cells requires knowledge of their antigen specificity. In experimental models, antigen-specific immune therapy, e.g., through induction of mucosal tolerance, is highly efficacious in experimental autoimmunity, particularly in models of arthritis. However, despite the rapidly growing knowledge base regarding the mechanisms of autoimmunity, translation of the findings from animal models into effective therapies in humans has been a major challenge (18, 19). Typical examples of this approach were the testing of tolerization to oral type II collagen and to glycoprotein-39 epitopes in clinical trials of RA (20–24). These trials can be considered the first deliberate attempts at antigen-specific immune therapy.

Despite the initial enthusiasm, the clinical results of these trials were not fully convincing, which is probably the consequence of several hurdles that could not be entirely taken into account at the time. Among such hurdles, it is worth mentioning that differences in protein glycosylation, due to the ongoing inflammatory process in RA and individual variations in processing of whole proteins, may influence T cell recognition of putative autoantigens, as has been demonstrated for type II collagen (25–27). This may, in turn, affect the efficacy of a therapeutic approach based on tolerization to proteins considered putative triggers of the disease. In addition, recent findings from a model using SKG mice and from our own studies (Roord S, et al: unpublished observations) highlight the concept that tolerization to disease-triggering antigens, if any can be identified, may be more difficult to achieve in the chronic phase of the disease (5, 28) if no attempt is made to intervene in those components of the inflammatory process dominated by macrophages and cytokine networks (8).

To date, most antigen-specific therapies have followed the traditional developmental path typical of the pharmaceutical industry. Thus, human studies have been designed to faithfully reproduce what was successfully found in animal models. This consolidated practice has been extremely successful for development of most of the current therapies, including first-generation biologic agents, and it is, understandably, the standard.

Clinical trials have therefore been designed in accordance with the concept that tolerization to an antigen demonstrated to be a trigger of arthritis in specific animal models leads to clinical control in such models, and that this control could be transposed to humans by the same approach. These assumptions may, unfortunately, be wrong, because human autoimmunity is extraordinarily more complex than that in animals, and it is likely that different and diverse antigen pathways are involved in the etiopathogenic events. By the time of intervention, the antigen, if any, that was the original trigger may be irrelevant to a self-sustaining process. Based on these previous experiences and our own observations, we suggest that, in the case of the developmental process involving antigen-specific therapy, the traditional animal-to-human sequential approach could be adapted.

Revisiting the trigger epitope concept in immunotherapy

The first conceptual change to consider may be that the search for the elusive trigger of human autoimmune disease, which has been focused on the antigen of choice for immunotherapy, may, instead, be focused on antigens that participate in the mechanisms of amplification and control of inflammation. An approach of this type may be fundamental in overcoming one of the major limitations of epitope-specific therapy in humans, i.e., the fact that the triggers are elusive and possibly multiple. More importantly, this approach may target mechanisms that are relevant and central to the perpetuation and modulation of the autoimmune inflammation generated by both the innate and the adaptive arms of the immune system, thus increasing the chances for the treatment to be clinically relevant.

Potential candidate antigens need to fulfill certain characteristics. In particular, candidate antigens should be present and possibly overexpressed at the site of inflammation. They should be immunologically relevant, thus triggering T cell responses and production of cytokines that contribute to modulating the inflammatory process locally and systemically. Self antigens with these characteristics often have molecular mimics whose crossrecognition may be necessary to initiate or modulate the immune reactivity. Several families of antigens with these characteristics have been described (29–31).

Heat-shock proteins (HSPs) are among the most notable of such proteins. In recent years, considerable progress has been made in characterizing the central role of HSPs in the modulation and amplification of inflammation, both in health and in disease (32–35). HSPs are evolutionary, highly conserved proteins that are present in the cells of virtually all living organisms and play essential roles in cell function. Expression of HSPs is up-regulated during conditions of cellular stress, including infection and inflammation, and as such they are a readily available antigenic challenge. Increased expression of endogenously produced Hsp60 and dnaJ has been documented at various sites of autoimmune inflammation (36–40).

Recognition of HSPs is, by default, perceived as a danger signal by the immune system, thus triggering a proinflammatory response that involves both the adaptive and the innate arms of the immune system. T cell proinflammatory responses to HSPs are found in several autoimmune diseases, including RA, juvenile idiopathic arthritis (JIA), juvenile diabetes mellitus, multiple sclerosis, and inflammatory bowel disease (34, 35, 41). HSPs are, however, not purely proinflammatory antigens. Years of research in disparate settings have, in fact, shown that recognition of HSPs often occurs in inflammatory chronic diseases and is often associated with a remitting form of such diseases. Therefore, responses to HSPs in autoimmunity seem to entail a certain degree of contradiction.

A recent important advance was achieved when individual epitopes, rather than whole proteins, were used as antigens. This approach enabled the identification, in several autoimmune diseases, of purely inflammatory epitopes (36, 38, 42–49) as well as epitopes with tolerogenic, or regulatory, characteristics, namely, inducers of production of tolerogenic cytokines and recognition of T cells with a regulatory phenotype (refs.50–52 and Massa M, et al: unpublished observations). Correlation between these responses and the clinical characteristics of patients with remitting–relapsing autoimmune diseases has helped to clarify the conundrum. In essence, HSPs appear to directly affect the pathogenesis of chronic autoimmune inflammation by contributing not only to its amplification and perpetuation, but also to its control through mechanisms predominantly associated with antigen-specific regulatory T cells. We have proposed that this mechanism be characterized as a “molecular dimmer” of inflammation (19, 38, 53).

As mentioned above, other antigens may also exhibit similar characteristics. Taken together, they would comprise a group of antigens that play a central role in autoimmune inflammation without necessarily being the triggers of it. Translated into therapy, an approach focused on this group of antigens would have considerable advantages, such as the possibility of acting on much broader and relevant pathogenic mechanisms, resulting in an increased likelihood of clinical success.

Consideration of biomarkers in trial design

As the direct outcome of years of evolution in molecular immunology, epitope-specific immunotherapy is one of the ideal testing fields to establish the relevance of biomarkers in clinical trial design and data interpretation. Several aspects of epitope-specific immunotherapy stem from our direct experience in RA.

We have found that the epitope to be used in immunotherapy needs to have demonstrated relevance in the population of patients to be tested. This approach has 2 advantages. First, it will allow for preselection of the population of patients to be enrolled in the trial, thus maximizing the chance for therapeutic success. Moreover, the use of biomarkers will identify an immunologic baseline. Treatment-induced immune deviation could then be considered as a secondary outcome. The identification of an immunologic baseline through this method of patient preselection would enable followup trials aimed at demonstrating the biologic effectiveness of the treatment.

In the early stages of clinical development (phase I/IIa), the use of immunologic biomarkers can significantly speed up the process, since fundamental information on the efficacy and safety of the biologic agents can be obtained. Although such markers are not being used at present, emerging new technology, such as tetramers, T cell capture, and the multiplex immunoassay, can be used to facilitate the development of specific tools for monitoring the joint-specific immune response (54–57). In further development, the use of immunologic biomarkers could be expanded to distinguish subpopulations with diverse response patterns and to identify surrogate markers of efficacy. This is a powerful, and yet underutilized, strategy for trial design and efficient decision making in clinical development, particularly in the early stages of disease.

The above-described approach is the direct outcome of our own experience. Recently, the results of a first phase I/II trial in RA in which direct immune modulation with an HSP peptide was used were published. The peptide used was dnaJP1, which has distinct proinflammatory properties in patients with RA (36, 38, 46). Following mucosal tolerization to dnaJP1 in a group of 15 RA patients, an intriguing change from proinflammatory to tolerogenic responses to dnaJP1 was observed, without noticeable side effects (45). Remarkably, the treatment did not affect the total number of dnaJP1-specific T cells, but rather led to a treatment-induced active immune deviation by regulatory T cells. A double-blind, placebo-controlled trial that has been designed along the same concepts is now under way to further demonstrate the clinical efficacy of this approach.

Objective of epitope-specific therapy: tolerization or restoration of regulatory T cell activity?

Another consequence of the shift of focus from triggering to pathogenic antigens in epitope-specific therapy is the possibility that “infectious tolerance” may develop, since regulatory T cell effects are being exploited on both the innate and the adaptive arms of the immune response. This new dimension to antigen-specific therapy comes from our increased understanding of how the immune system avoids reacting to itself. As T cells are activated during a normal immune response against an infectious agent, tissue damage often occurs, resulting in the release of self antigens from restricted sites. Given the plasticity of the TCR in engaging with MHC–peptide complexes on antigen-presenting cells (a single TCR potentially recognizes multiple MHC–peptide complexes) and the homology between self antigens and various microbial proteins, the immune system is facing a serious risk of autoreactivity and possible autoimmunity (58). Therefore, the immune system has a fail-safe mechanism that is dependent on a system of (counter)regulatory T cells known as Treg.

In animal models of experimental autoimmunity, Treg cells are essential for controlling inflammation, whereas depletion of these cells leads to unrestricted autoimmunity (59). Currently, at least 2 types of Treg cells have been described in humans: natural CD4,CD25^bright Treg cells, and induced Treg cells. Natural CD4,CD25^bright Treg cells are characterized by the elevated expression of CD25 (the IL-2 receptor chain) on the cell surface, as well as expression of the transcription factor foxp3 (60). Natural Treg cells are present both in healthy individuals and in individuals with autoimmune diseases, including patients with RA and patients with JIA (61–63). The other Treg cell type, induced Treg cells, or Tr1 cells, is antigen dependent and mainly characterized by the capacity to produce the regulatory cytokine IL-10 (64). The activation of induced Treg cells is partly controlled by the state of inflammation of the cells. For this reason, those cells have also been named anti-ergotype T cells, after the Greek ergon, which means work or activity (33). The immune system uses markers of activation, or ergotopes, which are antigens that act as beacons for Treg cells during the process of inflammation.

Previous studies by us and other investigators, using various different models, have shown that epitope-specific immunotherapy may act on epitope-specific Treg cells directly (32, 65–69). Interestingly, blockade of the TNFα pathway in RA patients may, at least temporarily, restore some of the abnormal Treg function in RA. These findings provide additional confirmation that combination therapy aimed at blockade of the TNFα pathway and blockade of epitopes may be a feasible approach. Indeed, in experimental arthritis, the combination of low-dose etanercept with epitope-specific immunotherapy enhanced treatment efficacy and induced Treg cells (Roord S, et al: unpublished observations).

Challenges of epitope-specific T cell therapy

As in all scientific fields in rapid evolution, epitope-specific T cell therapy needs initial validation. This will be provided by results of clinical and mechanistic studies that, while shedding additional light on the mechanisms, will also (hopefully) confirm their clinical relevance. Of particular importance in this context are the consistently exciting results obtained with different animal models when epitope-specific therapy and anticytokine therapy are combined. This approach typically leads to full disease control while significantly lowering the amount of anticytokine needed.

Another avenue to explore is the direct augmentation of responses toward HSP epitopes, which are naturally tolerogenic and contribute to down-regulation of inflammation. Much still needs to be done in this area. Future research should be aimed at defining novel antigen targets, and should focus on determining the optimal route and timing of immunotherapy.