The sentinel within: exploiting the immune system for cancer biomarkers

Abstract

The release of proteins from tumors triggers an immune response in cancer patients. These tumor antigens arise from several mechanisms including tumor-specific alterations in protein expression, mutation, folding, degradation, or intracellular localization. Responses to most tumor antigens are rarely observed in healthy individuals, making the response itself a biomarker that betrays the presence of underlying cancer. Antibody immune responses show promise as clinical biomarkers because antibodies have long half-lives in serum, are easy to measure, and are stable in blood samples. However, our understanding of the specificity and the impact of the immune response in early stages of cancer is limited. The immune response to cancer, whether endogenous or driven by vaccines, involves highly specific T lymphocytes (which target tumor-derived peptides bound to self-MHC proteins) and B lymphocytes (which generate antibodies to tumor-derived proteins). T cell target antigens have been identified either by expression cloning from tumor cDNA libraries, or by prediction based on patterns of antigen expression ("reverse immunology"). B cell targets have been similarly identified using the antibodies in patient sera to screen cDNA libraries derived from tumor cell lines. This review focuses on the application of recent advances in proteomics for the identification of tumor antigens. These advances are opening the door for targeted vaccine development, and for using immune response signatures as biomarkers for cancer diagnosis and monitoring.

Figure 1. The application of cancer proteomics to tumor immunology

The immune response to cancer, whether endogenous or driven by immunotherapy, involves a complex array of interactions between the tumor, host microenvironment, lymphocytes, antigen presenting cells, antibodies, cytokines and chemokines. Major arms of the immune system are shown in blue, and selected techniques for evaluation are shown in green. The identification of target antigens of B- and T-lymphocyte recognition may be identified on a proteome-wide basis, through cell fractionation, protein microarrays, nanospray mass spectrometry, and protein expression profiling of tumors in vivo. Cellular assays may use protein microarrays, microfluidics, and nanotechnology to identify target antigens as well as host responses. Microenvironmental interactions can be analyzed, as well as signal transduction pathways and costimulatory/regulatory molecules. This review will focus on B-cell and T-cell target antigen identification.

Figure 2. Classical model of antigen presentation

Intracellular antigens (i.e., viral) are cleaved into peptides by the proteasome and bound to MHC Class I molecules at the surface of cells for presentation to CD8⁺ T lymphocytes. Exogenous antigens are endocytosed by antigen presenting cells (APC), which is markedly enhanced by binding of serum antibodies both to antigen and to the Fcγ receptor on the APC, with subsequent degradation and binding of peptides to MHC Class II molecules for stimulation of CD4⁺ T lymphocytes. Because of this coordinated immune response, tumor antigens that cause antibody responses in patient sera may contain immunogenic T cell peptide epitopes as well (modified from D. Miklos).

Figure 3. The development of autoantibody biomarkers in cancer

The identification of target antigens using antibodies derived from patient sera is the first step in identification of potential biomarkers for cancer diagnosis or immune monitoring. In the field of autoimmunity, detailed evaluation of the autoantibody quantity (titration) and specificity has required development of a confirmatory ELISA using heterologous recombinant protein. Antigens that segregate patient sera from control sera (“informative antigens”) would be tested further. Analogous to DNA microarrays, well-annotated and blinded test sets and validation sets of sera (such as sets of sera from different stages or types of cancers) are required to identify areas of potential clinical applications. Finally, prospective clinical trial analysis is required to further validate the sensitivity, specificity, and predictive value of the assay. For tumor antigens, gap analysis that identifies false-negative sera (“missed cancers”) would then be further screened for additional autoantibody biomarkers. The potential clinical applications of these biomarkers include diagnosis, immune monitoring, vaccine development, and endpoint analysis of cancer therapeutics.

Figure 4. Methods of antigen identification

In traditional library screening (top row), cDNA libraries from tumor cells are expressed by phage, blotted on a membrane, and probed with patient sera. Confirmation of sensitivity and specificity requires recombinant expression of protein for ELISA analysis. Similar approaches of library screening may use phage display, yeast display or peptide libraries. Cell fractionation approaches (middle row) are a modification of the traditional western blot. Tumor cell lysates may be fractionated using microfluidic 2-dimensional or 3-dimensional separations, followed by printing of fractions on microarrays for probing with patient sera. Fractions identified in patient sera but not normal sera (“informative fractions”) are then further evaluated by mass spectrometry to identify target antigens, which then must be confirmed for sensitivity and specificity using ELISA or similar methods. In the bottom row, protein microarrays can be printed using recombinant proteins or proteins translated in situ, then probed with patient sera. Because the protein microarrays are addressable, no further antigen identification is required, and the sensitivity and specificity of multiple antigens may be evaluated in parallel with screening.

Figure 5. Identification of HLA-binding peptides by mass spectrometry

Peptide-MHC complexes are immunopurified from tumor cell (or antigen presenting cell, APC) lysates using antibodies directed against HLA molecules. Bound peptide mixtures are eluted in acid, and separated by nano-LC followed by MS/MS identification (top arrow). Alternatively, peptides may be directed applied to nanospray quadrupole TOF (bottom arrow), and predicted epitopes identified by fragmentation patterns.