New tools, new approaches and new ideas for vaccine development

Vaccine developers would do well to pay more attention to the explosion of information regarding T-cell epitopes and T-cell responses that is emerging from the nexus of bioinformatics, immunology and autoimmunity. Invasive pathogens may express thousands of proteins containing potentially millions of T-cell epitopes, yet a much smaller set of sequences plucked from this protein haystack interfaces with the cellular network that sets in motion events leading to immunity. The set of pathogen epitopes that interface with the host immune system is now known as the ‘immunome’. A number of recent papers point a clear path from the genome to the immunome, bringing us closer to understanding just what information about a pathogen is required for effective host immune defense. Indeed, we now have several approximations of how many epitopes are required to protect against infection. Similarly, the heterogeneity of host immune responses to T-cell epitopes responsible for producing immunity and self-tolerance is beginning to be better defined.

Exposure to foreign antigens induces the adaptive immune system to eliminate infection and develop protective immunity. Central to this effort are T lymphocytes, which operate as activators of B-lymphocyte growth and differentiation and as effector cells of cell-mediated immunity. T cells carry out these roles in response to peptide fragments of pathogen (or self) proteins that are displayed to them by major histocompatibility complex (MHC) proteins at the surface of antigen-presenting cells. One of the critical determinants of immunogenicity is the strength of peptide binding to MHC molecules [1]. Thus, methods to accurately model the MHC–peptide interface and accurately predict these immunogenic peptide sequences, also known as T-cell epitopes, have long been recognized as essential for mining protein sequence information for vaccine design. In the not-too-distant past, this task was accomplished by synthesizing overlapping peptides (usually 15 mers, overlapping by five amino acids) and measuring the T-cell response to each overlapping peptide. Given the availability of entire pathogen genomes for epitope mapping and the improved accuracy of T-cell epitope mapping tools, the overlapping peptide approach is no longer experimentally feasible or affordable for most researchers.

Fortunately, it is now easier to find T-cell epitopes that are central to T-cell activation and regulation thanks to the development of a wide range of in silico epitope-mapping tools over the past decade. The bioinformatics discipline that creates such mining tools is called immunoinformatics. Immunoinformatics tools significantly accelerate the discovery of peptide sequences that account for an overwhelmingly large proportion of T-cell responses to pathogens and self-proteins. Using these tools, not only are researchers able to identify new epitopes (and hence antigens) for the design of vaccines against pathogens but they also reveal the way that a host may sample and utilize a broader array of antigens than previously thought when it mounts an immune response.

The epitope-mapping trend was first established in the pregenomic era by Jay Berzofsky and colleagues at the US NIH, who mapped T-cell epitopes in malaria proteins and HIV proteins in the early 1980s. The arrival of the genomics age made it possible to leap-frog vaccine design, skipping the labor-intensive process of determining the structure and function of a protein prior to testing for protection against challenge. Rino Rappuoli of Chiron (now Novartis) has been one of the leading contributors to this approach, known as ‘reverse vaccinology’ [2]. Slightly different approaches have been pioneered by Alex Sette and colleagues [3] and the EpiVax Inc. team in Providence (RI, USA) [4], among others too numerous to mention here.

Unfortunately, for eager T-cell epitope mappers, not all epitope-mapping methods are equivalent. Complicating matters further, researchers often combine several tools. Moreover, while predictive tools may benefit from tightening, extrinsic factors, such as antigen abundance and processing, may still influence the final outcome of an immune response and are presently challenging to take into account at the predictive level. A ‘gold standard’ set of epitopes, useful for comparing and contrasting new tools against existing publicly available tools is now available on the Immune Epitope Database [101]. We expect that, as the field further matures, the means for validating, comparing and contrasting epitope prediction tools will improve.

An additional caveat relates to the discovery of a ‘new but old’ category of regulatory T cells (Tregs) that regulate immune responses (previously known as suppressor cells). These Tregs respond to epitopes in much the same way that T effector cells do, however, the result is quite different – suppression of the immune response. It is believed that the cells provide a ‘fail-safe’ mechanism for control of the immune response after clearance of the pathogen and also serve an important role in the control of autoimmunity. Tregs accomplish these tasks by secretion of anti-inflammatory cytokines, such as interleukin-10 and transforming growth factor-β. They can also lead to functional anergy in the circulating autoreactive B cells. The extent to which Tregs may be involved in the control of the immune response to pathogens is unknown but is now yielding to discovery with the advent of good T-cell epitope-mapping tools. Some pathogens are known to exploit the overlap between the regulation of T-cell and effector immune responses [5]. Thus, when searching for vaccine epitopes, vaccinologists are well advised to determine the phenotype of responding T cells, as effector T cells would generate a desirable vaccine response, whereas the Treg response would dampen the effect of the immunization.

One of the most remarkable findings that is emerging from the immunoinformatics approach to epitope mapping is that no single epitope is overwhelmingly immunodominant in terms of the host T-cell response to a pathogen [1,4,6]. In addition, epitope-driven approaches have led to the discovery that protective host immune responses are not confined to ‘virulence’ determinants but, rather, may involve T-cell response to a range of ‘housekeeping’, structural and functional proteins – and to proteins for which no known function has been identified. These discoveries led to the suggestion that we pay less heed to the concept of immunodominance in favor of a new view – that the immune system recognizes a broad range of pathogen-derived epitopes and that regulatory and structural proteins may also be relevant to a competent immune response. Since some ‘housekeeping’ proteins are often conserved among pathogen species, the concept of a ‘pan-genome-derived vaccine’ is now being explored.

In summary, scientists are now beginning to use T-cell epitope mapping tools to measure the breadth and overlap of pathogen immunomes that give rise to both favorable and unfavorable aspects of host immunity. They are also discovering the value of these tools in immunological applications beyond vaccine design that is leading to a rapid expansion of the immunoinformatics field. Potential applications of T-cell epitope-mapping tools include:

•	Comparison of pathogen immunomes: prior exposure to a particular pathogen may modulate immune responses to another pathogen, particularly if the epitopes overlap like two intersecting data sets in a Venn diagram [7]. Such epitopes can be eliminated (for pathogen-specific vaccines) or included (for pan-genome vaccines);
•	Measurement of the impact of pathogen exposure on autoimmunity: exposure to epitopes from parasites, such as Schistosoma mansoni, may diminish the chance of developing thyroid disease [8]. Likewise, careful evaluation of the presence of Treg immune responses to pathogen epitopes may improve vaccine design;
•	Assessment of the effect of priming on transplantation: exposure to heterologous T-cell epitopes from different pathogens may also prime cross-reactive T cells that impair tolerance to transplantation [9,10]. Transplantation success may be adversely affected;
•	Rational enhancement of immunogenicity: the half-life of class II MHC–peptide complexes appears to be the primary parameter that dictates the ultimate hierarchy of the T-cell response [1]. Modification of T-cell epitopes might improve the immune response to vaccines;
•	Modification of proteins to eliminate T-cell epitopes: some innovators are turning the vaccine paradigm on its head and using the epitope mapping tool to identify and eliminate T-cell epitopes in protein therapeutics [11,12]. New protein therapeutics that do not engender adverse immunogenicity may be developed.

Immunoinformatics is having a positive impact on vaccine development. New tools are leading to a new understanding of the host immune response and new approaches to developing vaccines. Researchers are now able to separate the immunological needle from the genome haystack, improving their likelihood of success when endeavoring to drive or diminish immune responses. In the course of that work, vaccinologists will bridge informatics and immunology, harnessing genome data, proteomics and immunology techniques in a new, interdisciplinary realm of inquiry that is bound to improve human health.