- Top of page
- Materials and Methods
- Disclosure Statement
Carcinoembryonic antigen (CEA) is a very common tumor marker because many types of solid cancer usually produce a variety of CEA and a highly sensitive measuring kit has been developed. However, immunological responses associated with CEA have not been fully characterized, and specifically a weak immunogenicity of CEA protein as a tumor antigen is reported in human leukocyte antigen (HLA)-A24-restricted CEA peptide-based cancer immunotherapy. These observations demonstrated that immunogenic and potent HLA-A24-restricted CTL epitope peptides derived from CEA protein are seemingly difficult to predict using a conventional bioinformatics approach based on primary amino acid sequence. In the present study, we developed an in silico docking simulation assay system of binding affinity between HLA-A24 protein and A24-restricted peptides using two software packages, AutoDock and MODELLER, and a crystal structure of HLA-A24 protein obtained from the Protein Data Bank. We compared the current assay system with HLA–peptide binding predictions of the bioinformatics and molecular analysis section (BIMAS) in terms of the prediction capability using MHC stabilization and peptide-stimulated CTL induction assays for CEA and other HLA-A24 peptides. The MHC stabilization score was inversely correlated with the affinity calculated in the docking simulation alone (r = −0.589, P = 0.015), not with BIMAS score or the IFN-γ production index. On the other hand, BIMAS was not significantly correlated with any other parameters. These results suggested that our in silico assay system has potential advantages in efficiency of epitope prediction over BIMAS and ease of use for bioinformaticians. (Cancer Sci 2011; 102: 690–696)
The carcinoembryonic antigen (CEA) is a very common tumor marker because many types of solid cancer usually produce a variety of CEA and a highly sensitive measuring kit has been developed.(1–3) However, immunological responses associated CEA have not been fully characterized except the observation that human leukocyte antigen (HLA)-A2 CEA epitope has been identified and contributed moderately to cancer vaccinations,(4,5) because there are several issues to be investigated. Specifically: (i) there are few efficient CEA epitopes with HLA-A24 restriction inducing cytotoxic T-lymphocytes (CTL) reaction in cancer patients; (ii) there have been few attempts to induce CEA-specific CTL using cancer patient-derived peripheral blood mononuclear cells (PBMC); and (iii) there have been few successful clinical trials using HLA-A24 restricted, which is well known as CEA peptide-based cancer vaccines.
Additionally, a weak immunogenicity or immunotolerance of CEA protein as a tumor antigen has been reported in cancer patients with advanced clinical stages.(6–8) Basically, CEA is originally expressed in the thymus and belongs to the CD66 family, which comprises highly homologous molecules expressed on hematopoietic cells, raising concerns about tolerance interfering with the development of anti-CEA immunity. Pickford et al.(6) reported that some effector T cells in PBMC can proliferate and some regulatory T cells can produce interleukin (IL)-10 with the stimulation by CEA protein using PBMC from 50 healthy volunteers. On the other hand, Crosti et al.(7) showed four CEA helper epitopes stimulating the proliferation of CD4+ T-lymphocytes from end-stage lung cancer patients. These results supported that CEA tolerance could be overcome by peptide vaccine recognized specifically by effector T cells.
Nukaya et al.(9) reported CEA652-A24-restricted peptide (TYACFVSNL) using healthy donor PBMC-based CTL induction assay for the first time and demonstrated that peptide-stimulated CTL efficiently killed CEA-positive cancer cell lines. The clinical trial by Matsuda et al.(10) and other researchers(11–13) using CEA652 peptide-treated dendritic cell (DC) vaccine showed that no significant clinical responses were seen, but some of the patients given vaccines developed in vitro CTL response against CEA peptide after vaccination. Considering that any HLA-A24 immunogenic CEA peptides besides CEA652 have not yet been found, it seems difficult to identify and predict immunogenic and potent HLA-A24-restricted CTL epitope peptides derived from CEA using conventional bioinformatics approach like HLA binding predictions of the bioinformatics and molecular analysis section (BIMAS) (http://www-bimas.cit.nih.gov/molbio/hla_bind/).
In the present study, we developed an in silico assay system to assess the affinity between HLA-A24 protein and A24-restricted peptide binder by combining two free software packages, protein 3D structure prediction software, MODELLER,(14) and protein-ligand docking simulation software, AutoDock.(15) In this method, we used a complex structure of HLA-A24 protein and its epitope peptide deposited in the Protein Data Bank (PDB)(16) as a template of 3D structure backbone. We then compared the current simulation assay with BIMAS in terms of the prediction capability and investigated the correlation of simulation results with other biological assays regarding HLA-A24 epitope binding.
- Top of page
- Materials and Methods
- Disclosure Statement
HLA molecules are cell surface proteins that bind to antigen peptides and present them as HLA–peptide complexes to T-lymphocytes, thereby inducing a specific immune response. The binding site of HLA-class I molecules is formed by a deep cleft between two α-helices on top of an extended β-sheet structure. Only approximately bound peptides with a high affinity could trigger a potent immune response. Therefore, to identify possible high-affinity peptides derived from tumor antigens is a very important theme for developing cancer vaccines that have therapeutic efficacy.
Recently, a large amount of data for crystal structures of HLA–peptide complexes has been accumulated, allowing bioinformaticians to predict molecular binding simulations.(20,21) However, despite biophysical and biochemical analysis of HLA complexes, accurate prediction of antigen peptide binding to HLA molecule is still difficult.
In the present study, we developed an in silico docking simulation assay between the HLA-A24 molecule and an antigen peptide based on computational structural analysis, aiming to predict optimal docking peptides with high affinity leading to effective CTL activation. Conventional algorithms based on primary amino acid sequence have been successful so far in predicting antigenic peptides binding to the HLA class I pocket, but such algorithms have limited accuracy and provide no structural information.(22,23)
Recently, the development of algorithms to predict the structure of MHC–peptide complex has been tried. Briefly, Logean and Rognan(24) used a binding free energy scoring function in a newly developed EpiDock, and Desmet et al.(25) developed a quantitative structure-based affinity scoring method for prediction of anchoring peptide side chains in peptide–HLA complex.
Unfortunately, these methods have not been applied to the intact HLA-A24 protein because of no such available structure in PDB at that time. We performed, for the first time, the epitope prediction for HLA-A24 protein using the 3D complex structure of the protein and epitope (PDB code 2BCK). In this study, we applied two software packages, AutoDock and MODELLER, to a novel in silico assay system of peptide binding affinity. MODELLER was performed to build the initial conformation of any given peptide, and AutoDock was used to assess the affinity (ΔG) between HLA-A24 protein and the given peptide. In general, AutoDock is originally available for predicting the binding mode of low-molecular compounds in the context of receptor–ligand docking, and is not always suitable for highly flexible ligands such as 9-mer peptides. In the case of HLA–peptide docking, however, the binding pocket and epitope conformation have been well investigated, and the conformation search parameters in simulation can be limited or narrowed down differently from so called “blind docking.” We have shown that a local search simulation around the binding pocket of HLA-A24 protein reproduced the affinities between the protein and A24-restricted peptides. More importantly, all of these procedures are performed by combination of free software (AutoDock and MODELLER), suggesting that our in silico simulation assay is easily available to many bioinformaticians with the same interests. Recently, Fuhrmann et al.(26) reported a receptor–ligand docking simulation method based on a new Lamarckian genetic algorithm which can treat a large number of degrees of freedom.(26) Thus, the further improvement of an in silico binding assay system may be possible for not only HLA class I epitope peptides (approximately 10-mer) but also class II epitope peptides (approximately 20-mer). Additionally, we have performed MHC stabilization and CTL induction assays regarding various CEA and other HLA-A24-restricted peptides and investigated the correlation of docking simulation assay with other immunological parameters, the MHC stabilization score and IFN-γ production index.
Cytotoxic T-lymphocyte induction experiments using PBMC from CEA-positive cancer patients demonstrated that CEA A24-6 and 9 were mainly positive for CTL inducing activity. These peptides were shown to have moderate to high affinity in a docking simulation assay. These observations may support the negative selection theory that intermediate binders can be sometimes good CTL inducers in cancer patients because CTL precursors responding to high affinity peptides have been depleted in the thymus by negative selection, and such CTL may be exhausted after exposure to specific cancer antigen peptides.(27)
Unfortunately, CEA A24-3, which has the highest affinity in a docking assay, did not show strong CTL activation. Similarly, CEA A24-1, with the highest affinity in MHC stabilization, showed marginal binding score in a docking simulation, which indicates the discrepancy of affinity between docking simulations and other immunological binding assays. In our current method, unlike peptide epitopes, we did not permit the flexibility of the epitope-binding pocket structure of the HLA protein because of long computational time. The MHC stability assay suggests that CEA A24-1 is a good binder (Fig. 1), but its docking affinity is not high (Fig. 3). Since CEA A24-1 is the second largest peptide (1082.2 Da) among those in Table 1, it might not stably occupy a rigid binding pocket of HLA-A24 protein in a docking simulation. Actually, CEA A24-7, the largest peptide (1136.2 Da), has shown the lowest docking affinity. Although a strong correlation was not observed between the epitope’s molecular weight and docking affinity (r2 = 0.16), the rigid pocket simulation condition for HLA protein may have affected the docking conformation and affinity. A docking simulation that allows a pocket structure to be flexible is a challenge for the future, and it may be solved by advances in computer technology.
Most interestingly, CEA652 peptide, which is well known for specific CTL epitopes, did not show any positive CTL inducing activity in all patient PBMC. On the other hand, CEA A24-6, which showed the highest frequency of CTL induction, has not been reported as a CTL epitope previously, but few CTL induction studies using CEA-positive cancer patient PBMC have been done. Therefore, there might be some difference in CTL culture environments including immunological regulation between healthy donor and cancer patients, resulting in selection of antigenic peptides. Crosti et al.(7) showed four CEA helper epitopes stimulating the proliferation of CD4+ T-lymphocytes from end-stage lung cancer patients. These results supported positively our results that CTL inducing activity for any CEA peptides with HLA-A24 restriction was seen in 12 of 29 cancer patients.
Nukaya et al.(9) reported CEA652-A24-restricted peptide using healthy donor PBMC-based CTL induction assay for the first time and demonstrated peptide-stimulated CTL efficiently killed CEA-positive cancer cell lines. The CEA652 peptide has been also used for DC-based cancer vaccine against CEA-positive gastrointestinal solid cancers. The clinical trial by Matsuda et al.(10) using CEA-peptide treated DC vaccine showed that no significant clinical responses were seen, but four of seven patients developed in vitro CTL response against CEA peptide after vaccination. In the present study 44% of cancer patients demonstrated CTL activity against various CEA peptides; however, there was no response to CEA652 peptide.
In another CTL induction assay using patient PBMC and 16 well-known HLA-A24 peptides including CEA peptides, six peptides were positive for specific CTL activation. With regard to these 16 peptides, we also obtained other immunological data. In Figure 5, correlations between HLA-A24 peptide docking simulation and other immunological parameters, including BIMAS, MHC stabilization and IFN-γ production, were investigated using a correlation analysis. The binding affinity predicted in docking simulation was inversely correlated with MHC stabilization score (r = −0.589, P = 0.015), whereas BIMAS was not significantly correlated with that or any other parameter. These observations suggested that our docking simulation assay might have potential advantage in efficiency of epitope prediction over BIMAS. However, our docking simulation assay is still not developed enough to compare with other predictive programs precisely, and the accumulation of more simulation data is needed.
Recently, novel conventional algorithms based on primary amino acid sequence like BIMAS have been developed. They show better prediction than BIMAS,(28–30) but such algorithms have limited accuracy. Similarly, our in silico assay system has yet to predict genuine CTL epitope peptides with high affinity. Computational methods for modeling of peptide-HLA binding include threading, molecular dynamics and approaches based on HLA binding pocket recognition. Technical improvements for all methods should be requested to elevate the accuracy of CTL epitope prediction in future.