Purpose: Cyclin D1, a key cell cycle regulator, is overexpressed in multiple types of cancer. Such tumor-associated genes may be useful targets for cancer immunotherapy. Nevertheless, it had previously been suggested that efficient T cells recognizing cyclin D1-derived epitopes are absent from the repertoire because of thymic deletion. We attempted to induce autologous CTL from healthy donors and patients with cyclin D1-overexpressing tumors using a highly efficient T-cell expansion system based on CD40-activated B cells as antigen-presenting cells.
Experimental Design: Cyclin D1-derived, HLA-A*0201–restricted epitopes were predicted by multiple computer algorithms, screened in HLA-A2-binding assays, and used for T-cell stimulation. The generated CTL lines and clones were analyzed by IFN-γ enzyme-linked immunosorbent spot assay or cytolysis assay.
Results: After screening, at least two naturally processed and presented HLA-A*0201–binding cyclin D1 epitopes were identified. CTL specific for these epitopes could be successfully generated from HLA-A2+ donors. T cells efficiently recognized target cells pulsed with the cognate peptide and cyclin D1-expressing tumor cell lines in an HLA-A*0201–restricted manner. More importantly, HLA-A*0201–matched, primary cyclin D1+ tumor cells were efficiently recognized by cyclin D1-specific CTL. These CTL could be generated from patients with mantle cell lymphoma and cyclin D1+ colon cancer.
Conclusions: These results underscore that cyclin D1 needs to be considered as a target for broad-based antitumor immunotherapy.
Cyclin D1 (CycD1) has previously been studied as a potential tumor antigen because of its essential role in tumorigenesis and overexpression in multiple types of cancer. However, it had been concluded that autoreactive CTL are not present in the repertoire because of thymic deletion. Therefore, immune intervention was only considered possible in a highly select allogeneic setting. Using CD40-activated B cells as highly efficient antigen-presenting cells, we were able to generate in an autologous setting, even in patients with cancer, CycD1-specific CTL that lyse tumor cells endogenously expressing CycD1. This indicates that CycD1 is a promising target for active and passive immunotherapy in multiple tumor histologies.
Cell cycle deregulation is one of the hallmarks of cancer (1, 2). Several cyclin genes have been shown to be overexpressed in cancer, and cell cycle regulators have been suggested as potential targets of antitumor immune intervention (3–6).
Cyclin D1 is a cell cycle regulator that is crucial for the G1-S transition by interacting with several cyclin-dependent kinases (2, 7). Cyclin D1 is a proto-oncogene and has been shown to be overexpressed in major malignancies, including colorectal, gastric, esophageal, lung, kidney, and breast cancer, as well as leukemia and lymphoma with little expression in normal tissue (8–12). Cyclin D1, therefore, has been long under consideration as a tumor antigen and a target for anticancer immune intervention (6, 11–13). Nevertheless, similar to other overexpressed genes, it has been difficult to generate autologous CTL to cyclin D1 (1, 6). It has therefore been concluded that the human and murine immune systems are tolerant to the thymically expressed autoantigen cyclin D1, and an efficient immune response cannot be elicited (13). This obstacle has elegantly been circumvented by generating cyclin D1-specific allogeneic CTL from HLA-A*0201− donors that recognize the LLGATCMFV (CycD1_101-9) peptide in the context of HLA-A*0201 and efficiently lyse cyclin D1+ tumor cells, showing that this epitope is naturally processed and presented (1, 13). These results underline the potential of cyclin D1 as a target for immunotherapy.
It has been shown that autologous, CD40-activated B (CD40-B) cells efficiently present antigen, chemoattract T cells in vitro, and can be used to expand memory and prime naïve CD4+ and CD8+ T cells in vitro (14–18). Using this system, we have previously shown that the expansion of T cells specific for epitopes known to be processed and presented is readily possible for several self-antigens (15, 19–23). Furthermore, we have observed that autoantigen-specific T cells can be expanded from the peripheral blood of healthy donors and patients with cancer that had previously been considered absent from the repertoire (21).9
von Bergwelt-Baildon, in preparation.
We therefore assessed whether autologous cyclin D1-specific CTL can be detected using CD40-B cells as antigen-presenting cells in a reverse immunology approach.
Materials and Methods
Donors and patient samples. Peripheral blood from healthy volunteers and patients was obtained by phlebotomy following informed consent and approval by the institutional review board. Peripheral blood mononuclear cells were purified by Ficoll (Amersham Biosciences) density centrifugation, and CD8+ cells were separated using CD8 MicroBeads (Miltenyi Biotec) and cryopreserved until use.
Peptides and peptide prediction. Cyclin D1 or other tumor antigen–derived peptides (Table 1) were purchased from Sigma Genosys Biotechnologies and New England Peptides. Binding of peptides to HLA-A*0201 was predicted using four publicly available algorithms: SYFPEITHI, BIMAS, LPpep, and MHCPred (24–27). The peptides were ranked for each algorithm and sorted by a cumulative score.
|Start position .||Sequence .||BIMAS .||LPpep .||SYFPEITHI .||MHCPred .||T2 binding .||FI* .|
|Start position .||Sequence .||BIMAS .||LPpep .||SYFPEITHI .||MHCPred .||T2 binding .||FI* .|
Abbreviations: FI, fluorescence index; n.a., not applicable.
Binding affinity shown as fluorescence index [(mean fluorescence intensity)peptide pulsed T2 / (mean fluorescence intensity)unpulsed T2] in the T2 binding assay.
HLA-A*0201 peptide binding and complex stability assay. Following previously described methods, peptide binding was assayed using T2 cells (28). An increase of HLA-A*0201 expression on T2 cells reflects stabilization of MHC complexes by the addition of exogenous peptides and was quantified using the fluorescence index (mean fluorescence intensity with peptide / mean fluorescence intensity without peptide).
Cell lines. The cell lines T2, U266, K562, OVCAR, and DU145 were obtained from American Type Culture Collection, and SK-MM2 and OPM-2 were from Dr. R. Schmitz (University of Essen, Essen, Germany). CD40-B cells were generated as previously described (15). Lymphoblastoid cell lines were established from the CD40-B cells with supernatant of an EBV-producing cell line (B95-8, American Type Culture Collection) in RPMI 1640 (Invitrogen) supplemented with 10% FCS (Invitrogen).
Generation of CTL lines and cytotoxicity assay. CD40-B cells were incubated with peptide, irradiated (32 Gy), and added to autologous peripheral blood mononuclear cells or purified CD8+ T cells in AIM-V (Invitrogen) containing 5% pooled human serum, glutamine, gentamicin, and interleukin-7 (10 ng/mL; Endogen). On day 7, T-cell cultures were harvested, washed, and restimulated with fresh peptide-pulsed CD40-B cells and interleukin-7. This was repeated on days 14, 21, and 28. Interleukin-2 (30 IU/mL; Chiron) was introduced into the cultures at day 8 and every 3 to 4 d thereafter.
CTL lines were tested for cytotoxicity using standard 4-h 51Cr-release assay (15) or 2-h Eu-release assay (29). In brief, targets were labeled with 51Cr or Eu, and 5 × 103 labeled cells per well were plated with various concentrations of effector cells. The percentage of cytotoxicity was calculated as follows: [(experiment − spontaneous release) / (maximum release − spontaneous release)] × 100.
IFN-γ enzyme-linked immunosorbent spot assay. Enzyme-linked immunosorbent spot (ELISPOT) analysis for IFN-γ secretion using human CTL was carried out as previously described (20). The spots were counted using an automated ELISPOT reader (Aelvis).
Identification of HLA-A*0201–binding peptides. Candidate epitopes from the cell cycle regulator cyclin D1 were identified using four computational epitope prediction algorithms (SYFPEITHI, BIMAS, MHCPred, and LPpep). All candidate epitopes were ranked according to the sum of the individual prediction's rank. Twelve nonameric epitopes were identified to potentially bind highly to HLA-A*0201 (Table 1). A cellular binding assay was done to validate the predicted binding capacity. Of the 12 candidates, 5 nonameric peptides bound with at least weak affinity (fluorescence index >2; Table 1).
The CycD1_101-9 epitope (LLGATCMFV) had previously been shown to be naturally processed and presented but only be recognized by allogeneic HLA-A*0201–restricted CTL (1). We show here that these epitopes bind efficiently to HLA-A*0201. Interestingly, an almost identical decamer epitope (CycD1_101-10; LLGATCMFVA) was also predicted to bind to HLA-A*0201 (Table 1), and indeed, we were able to show significant binding of this epitope.
Expansion of peptide-specific CTL using CD40-B cells as antigen-presenting cells. We used a highly efficient autologous T-cell expansion system that had been shown to prime naïve T cells to determine whether CTL specific for the identified epitopes are present in the repertoire. Within the HLA-A*0201–binding peptides we identified, the two nonameric peptides with a high BIMAS score (CycD1_101-9, CycD1_228; RLTRFLSRV) and one derivative decamer (CycD1_101-10); were chosen as first candidates for generation of antigen-specific CTL. CTL specific for these peptides were readily generated from the peripheral blood of healthy HLA-A*0201+ volunteers by priming highly purified CD3+CD8+CD16−CD56− T cells followed by weekly restimulations for 3 to 4 weeks with peptide-pulsed autologous CD40-B cells. These CTL efficiently lysed T2 or CD40-B cells pulsed with the cognate peptide but did not lyse unpulsed T2 or CD40-B cells (Fig. 1A). Target cells loaded with an irrelevant peptide were also not lysed by the CTL (data not shown). These CycD1_101-9–specific CTL not only recognized the CycD1_101-9 peptide but also the CycD1_101-10 peptide–pulsing target cells (Fig. 1B), and vice versa, CTL generated against the CycD1_101-10 recognized cells pulsed with the CycD1_101-9 (data not shown).
We next addressed to what extent CTL recognizing the CycD1_101-9, CycD1_101-10, or CycD1_228 peptide in context of HLA-A*0201 are present in the repertoire of healthy individuals. Figure 1C shows 13 successfully expanded cyclin D1-specific CTL lines lysing 14% to 98% of the relevant peptide loaded T2 cells. We have also observed other 10 independent CTL lines stimulated with CycD1_101-9–pulsed CD40-B cells recognized the relevant peptide-pulsed target cells specifically by IFN-γ ELISPOT assay (data not shown). In summary, we have successfully established autologous cyclin D1-specific CTL lines from 9 of 10 HLA-A*0201+ donors against the 3 epitopes (Fig. 1D), showing that the T cells specific for cyclin D1 are still preserved in the peripheral T-cell repertoire during adult life.
Endogenous processing and presentation by tumor cells. To determine if the above epitopes are actually processed and presented by tumor cells and whether CTL recognize the naturally processed epitopes, HLA-A*0201+ and HLA-A*0201− tumor cell lines were tested for susceptibility to lysis by CycD1_101-9–, CycD1_101-10–, and CycD1_228-specific CTL. HLA-A*0201+ multiple myeloma cell line U266 (cyclin D1+) was efficiently lysed by CTL specific for all three epitopes, whereas the HLA-A*0201− ovarian carcinoma line OVCAR (HLA-A*0201−; cyclin D1+) and prostate cell line DU145 (HLA-A*0201−; cyclin D1+) were only killed at background levels (Fig. 2). Furthermore, autologous HLA-A*0201+ CD40-B cells (cyclin D1−) were equally not killed, showing that CTL specific for epitopes naturally processed and presented by CD40-B cells during T-cell expansion were not significantly expanded (Fig. 2, bottom). As observed for other tumor antigens, killing varied depending on time point and donor between high efficiency but high background killing (Fig. 2, top) and lower efficiency but higher specificity (Fig. 2, bottom).
Recognition of primary tumor cells by cyclin D1–specific CTL. To further investigate the potential of cyclin D1 as tumor antigen, we cloned CTL from bulk CD8+ T-cell cultures stimulated by autologous CD40-B cells pulsed with CycD1_101-9 peptide by limiting dilution. In an ELISPOT assay, these CTL clones showed the same specificity as that shown by the parent bulk CTL line through recognition of peptide-pulsed T2 cells (Fig. 3A). The specificity of one of the CTL clone, 2-9, was further confirmed by Eu-release cytotoxicity assay. In this experimental approach, CTL clone 2-9 efficiently lysed T2 cells pulsed with CycD1_101-9, U266 cells, and to a lesser extent, plasma cell leukemia cell line SK-MM-2 (HLA-A*0201+; cyclin D1+), whereas myeloma cell line OPM-2 (HLA-A*0201+; cyclin D1−) and irrelevant peptide-pulsed T2 cells were not lysed (Fig. 3B). Furthermore, cytolysis was inhibited by anti–MHC class I blockade and no natural killer activity was detected in cytotoxicity assays using K562 cells as targets (data not shown).
Next, we attempted to investigate whether primary tumor cells are recognized by the CycD1_101-9–specific CTL clone. Single-cell suspension of lymph nodes from three patients with mantle cell lymphoma and peripheral blood mononuclear cells from a patient with plasma cell leukemia with t(11, 14) chromosomal translocation were tested in IFN-γ ELISPOT assay. Cyclin D1 expression in all samples was confirmed by immunohistochemistry (data not shown). As shown in Fig. 3C, these primary tumor samples were recognized by the CycD1_101-9–specific CTL clone 2-9 in an HLA-A*0201 restricted manner. These results suggest that cyclin D1-specific cellular immunity can be generated, which has potential to successfully target cyclin D1-expressing tumor cells.
Specific CTL in patients with cancer with cyclin D1-expressing tumors. To address whether cyclin D1-specific cellular immunity can be induced in patients with cancer or whether this is prohibited by mechanisms such as peripheral tolerance, we attempted to generate CycD1_101-9– or CycD1_228-specific CTL in patients with mantle cell lymphoma, which is defined as a cyclin D1-overexpressing tumor, and cyclin D1+ colon cancer. Magnetically separated CD8+ cells from three HLA-A*0201+ patients with mantle cell lymphoma were stimulated with autologous CD40-B cells pulsed with CycD1_101-9 or CycD1_228 peptide. After four to five weekly stimulations, bulk CTL lines were tested in IFN-γ ELISPOT assay. Cyclin D1-specific CTL lines were successfully generated from two of three patients (CycD1_101-9–specific CTL in two patients and CycD1_228-specific CTL in one patient; Fig. 4A). T cells from patient 2 did not respond to antigen-specific stimulation. Experiments to address whether this was because of a general antigen-presenting cell or T-cell defect in this patient with advanced-stage cancer were not done because of the limited amounts of peripheral blood mononuclear cells. Similarly, cyclin D1-specific CTL were successfully generated from four of four patients with cyclin D1-positive colon cancer (CycD1_101-9–specific CTL in two patients and CycD1_228-specific CTL in three patients; Fig. 4B and C). This shows that cyclin D1-specific cellular immunity persists even in patients with cancer.
Here we show the successful expansion of autologous cyclin D1-specific CTL in 9 of 10 HLA-A*0201+ individuals, 2 of 3 patients with mantle cell lymphoma, and 4 of 4 patients with colon cancer. We identified a total of three epitopes for which specific autologous CTL were generated in the context of HLA-A*0201. Furthermore, using HLA-A*0201+ and HLA-A*0201− target cells, we show that the peptides CycD1_228 and CycD1_101 are naturally processed and presented and that T cells generated against these epitopes are of sufficient efficiency to lyse cyclin D1-expressing tumor cells. Of note, CTL specific for the peptides CycD1_22, CycD1_195, and CycD1_204 were also generated in two of two donors.10
Kendo E, Gyschok L, Schultze JL, von Bergwelt-Baildon MS, unpublished results.
These results differ from previous observations showing that a peptide derived from murine cyclin D1 (cyc20) does bind to Db and stimulate high-avidity T cells that fail to recognize target cells expressing the gene endogenously. Other MHC class I–binding peptides induced only low avidity (cyc41) or no CTL (cyc84, 147, 181). Considering the thymic expression of cyclin D1, these findings were explained by an absence of specific CTL from the repertoire (1, 6, 13). As recently reported, medullary thymic epithelial cells promiscuously express tissue-specific antigens (including tumor-specific antigens such as cancer-testis antigen; refs. 30, 31); thus, T cells specific for these antigens could be deleted by negative selection in thymus. However, several researchers have reported this tumor antigen–specific CTL generation in an autologous setting, suggesting that negative selection could not delete self-antigen–specific T cells completely. Whereas the frequency of these antigen-specific T cells might be varied from antigen to antigen, it seems to be possible to expand self-antigen–specific T cells using an efficient T-cell expansion system. In previous reports about cyclin D1-specific T cells, most likely the used antigen-presenting cells (T2, C1R-A2, murine splenocytes) are less potent than CD40-B cells and thus unable to expand low-frequency precursor populations. These findings are further supported by a recent report showing that 11.3% of patients with prostate cancer generate antibodies (IgG) to cyclin D1 (32). This further underlines the notion that cyclin D1-dependent immunity could be induced in patients with cancer.
Recently, regulatory T cells, especially CD4+25+ T cells, have been intensively investigated and are revealed to play a key role in peripheral tolerance (33). The depletion or absence of regulatory T cells by anti-CD25 antibody in mice (34) or patients with immune dysregulation, polyendocrinopathy, enteropathy, or X-linked syndrome causes severe autoimmune disease (35), indicating that central tolerance (thymic deletion) does not eliminate self-antigen–specific T cells completely (36). Therefore, a fraction of self-antigen–specific T cells seems to be able to leave the thymus and exist in the periphery. It is consistent that, in this study, CTL specific for cyclin D1 could be generated from multiple HLA-A*0201–positive individuals in an autologous setting. Based on these findings, it seems possible to generate CTL specific for other self-antigens, at least in an in vitro setting.
Several other cyclins (e.g., cyclin A, cyclin B1, and cyclin E) have been shown to be overexpressed in multiple types of cancer. Cyclin B1–specific antibodies have been shown in hepatocellular carcinoma, prostate, lung, colorectal, and breast cancer, and MHC class I–binding peptides have been identified that stimulate CTL from patients with cancer (5, 37–39). In view of these encouraging results, an increasing number of studies are currently addressing the role of the cyclins as tumor antigens in immunosurveillance and immunotherapy (40).
Taken together, the above data underscore that cyclin D1 needs to be considered as a target for broad-based antitumor immune intervention.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: The Mildred Scheel Scholarship and Max-Eder Young Investigator Award from Deutsche Krebshilfe (M.S. von Bergwelt-Baildon), the Alexander von Humboldt Stiftung and Kanae Foundation for Life and Socio-Medical Science (E. Kondo), the Deutsche Forschungsgemeinschaft and Multiple Myeloma Research Foundation (B. Maecker), NIH grants P01-CA-66996 and P01-CA-78378 (L.M. Nadler), and a Sofja Kovalevskaja Award by the Alexander von Humboldt Foundation and a Translational Research Award by the Leukemia and Lymphoma Society (J.L. Schultze).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. R. Schmitz for providing us with the SK-MM2 and OPM-2 cell lines, and Dr. Gordon Freeman for providing the NIH3T3 human CD40L cells.