Induction of Bcl-xL-Specific Cytotoxic T Lymphocytes in Mice


  • H. L. Larsen,

    1. Department of International Health, Immunology and Microbiology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • M. H. Andersen,

    1. Department of Hematology, Center for Cancer Immune Therapy (CCIT), Herlev University Hospital, Herlev, Denmark
    Search for more papers by this author
  • H. H. Wandall,

    1. Department of Cellular and Molecular Medicine, Copenhagen Center for Glycomics, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • C. B. Madsen,

    1. Department of International Health, Immunology and Microbiology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    2. Department of Cellular and Molecular Medicine, Copenhagen Center for Glycomics, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • R. E. Christensen,

    1. Department of International Health, Immunology and Microbiology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • T. R. Petersen,

    1. Malaghan Institute of Medical Research, Victoria University, Wellington, New Zealand
    Search for more papers by this author
  • A. E. Pedersen

    Corresponding author
    1. Department of International Health, Immunology and Microbiology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    • Correspondence to: A. E. Pedersen, Department of International Health, Immunology and Microbiology, Faculty of Health and Medical Sciences, University of Copenhagen, Blegdamsvej 3, DK-2200 Copenhagen N, Denmark. E-mail:

    Search for more papers by this author


The induction of active immunity against tumour-associated antigens to prevent relapse of cancer is a promising approach but has so far shown only low efficacy. This low efficacy may in part be due to clonal escape of tumour cell variants by the downregulation of antigen expression or inflammation-induced dedifferentiation. Identification of novel tumour-associated antigens that at the same time are essential for continued tumour cell survival is thus critical for the development of active cancer vaccinations. At the same time, identification of novel endogenous murine tumour antigens will help improve preclinical development of cancer immunotherapy. The anti-apoptotic protein Bcl-xL has been suggested to be such an essential tumour antigen, but the lack of well-defined murine epitopes have delayed preclinical studies of Bcl-xL-targeting cancer vaccines. Here, we report the identification of two novel murine tumour-associated epitopes TAYQSFEQV and AFFSFGGAL derived from mouse Bcl-xL. Dendritic cell (DC)-based vaccination induced CD8+ T cells capable of producing IFN-γ upon restimulation with these epitopes. Thus, our data may benefit the design of future immunotherapy strategies by providing a preclinical model for cancer vaccination with an endogenous tumour antigen that can be combined with other cancer treatments.


Cancer immunotherapy, with the use of unspecific immunotherapy like Interleukin-2 and Interferon-α and more recently CTLA4 blockade (Ipilimumab®) [1] and anti-PD-1/PD-L1 antibodies [2, 3], has demonstrated efficacy in particular in malignant melanoma. Also, it is believed that the recent success of a dendritic cell–like vaccine against metastatic prostate cancer [4] will lead to breakthroughs in cancer vaccination for other malignancies.

For future development of cancer vaccines and immunotherapy, establishment of preclinical testing in mouse models is essential. Reliable mouse models of cancer immunotherapy are dependent on the use of endogenous tumour antigens. However, so far, several studies have used model antigens like ovalbumine-expressing B16 melanoma cells [5-7], whereas relatively few endogenous tumour antigens have been characterized and tested in mouse models. Most studies with B16 cells have used gp100 [7] or TRP2 [8, 9] as standard endogenous tumour antigens, whereas many recently defined tumour antigens are less immunogenic [10]. Thus, characterization of new and potent MHC class-I-binding CD8+ T-cell epitopes is essential for further preclinical cancer vaccine development.

Cancer vaccination is believed to be optima when it relies on immunization against tumour-associated antigens that are essential for continuous tumour cell survival and therefore not likely to be downregulated by the tumour during immune-selective pressure. Anti-apoptotic proteins, which prevent execution of the apoptotic programme in tumour cells, may represent such essential proteins. Here, anti-apoptotic proteins of the Bcl-2 family members could be important targets as upregulation of such is a common tumour mechanism to ensure tumour cell survival despite genomic instability and a hostile tumour microenvironment, thus facilitating tumour progression, metastasis and resistance to chemotherapy and radiotherapy [11].

Bcl-xL (Bcl2l1) is a Bcl-2 family member. The Bcl-X gene gives rise to two splice variants; the longer Bcl-xL and the shorter product Bcl-xS in which 62 amino acids at position 126-188 of Bcl-xL are absent due to alternative splicing [12]. Bcl-xS is pro-apoptotic, whereas Bcl-xL is an inhibitor of the intrinsic apoptosis pathway. Bcl-xL is localized to the mitochondrial outer membrane and exerts its anti-apoptotic function through a yet undefined mechanism, involving formation of heterodimeric complexes with other Bcl-2 family members that prevents permeabilization of the mitochondrial outer membrane [13]. Increased expression of Bcl-xL has been observed in various haematological cancers and solid cancers such as breast cancer, pancreatic cancer and melanoma and is associated with poor prognosis and decreased response to chemotherapy and radiotherapy [14]. Thus, in particular, the 62 AA sequence, which is absent in Bcl-xS, may be a potential tumour antigen in cancer vaccination because this sequence is not constitutively expressed and at the same time may be overexpressed in tumours, thus enabling break of immunological tolerance. Indeed, spontaneous CD8+ T-cell responses against human Bcl-xL-derived epitopes have been identified in patients with cancer but not in healthy controls [15], and these Bcl-xL-specific T cells have been shown to be capable of lysing tumour cells pulsed with the cognate Bcl-xL-derived peptides or tumour cells with high endogenous Bcl-xL expression [16].

In the present study, we investigated the presence of the anti-apoptotic proteins Bcl-xL, XIAP, survivin and Bcl-2 in common cell lines used for tumour challenge studies in mice. Based on expression data and in silico prediction for H-2 binding, the putative H-2Kb binders TAYQSFEQV (AA118-126) and AFFSFGGAL (AA143-151) from murine Bcl-xL were selected as candidate epitopes for the induction of specific CD8+ T-cell responses. DC-based vaccination with these Bcl-xL-derived epitopes were able to induce a substantial number of Bcl-xL-reactive IFN-γ-producing CD8+ T cells.

Materials and methods


Six to eight weeks old female C57Bl/6 mice (Taconic, Ry, Denmark) were housed at The Department of Experimental Medicine, University of Copenhagen under controlled microbial conditions.

Cell lines

Cell lines used were EL4 and B16. All cell lines were cultured in 75-cm2 culture flasks (NUNC, Roskilde, Denmark) in culture medium (CM) consisting of RPMI-1640 supplemented with NaHCO3, GlutamaxTM, 10% heat inactivated foetal calf serum (FCS) (Sera Laboratories International Ltd, West Sussex, UK) and 1% penicillin/streptomycin. Adherent cells were detached by treatment with trypsin-EDTA (GIBCO). When relevant, cells were stimulated with 20 ng/ml IFN-γ for 24 h.

Epitope prediction

The amino acid sequence corresponding to mouse Bcl-xL was obtained from the UniProt database ( Relevant amino acid sequences were pasted into the in silico epitope prediction algorithm NetMHC3.2 (, and MHC class-I-binding epitopes were predicted by scrutinizing the amino acid sequences for 8-11-mer peptides with predicted affinities to C57Bl/6 alleles (H-2b).

Peptide synthesis

MHC class-I-binding peptides used in vaccination studies were synthesized by Schafer-N (Copenhagen, Denmark) and included SIINFEKL from ovalbumin and TAYQSFEQV + AFFSFGGAL from mouse Bcl-xL. The MHC class-II-binding peptide HepCore, derived from hepatitis B core antigen was used as a CD4+ helper peptide where indicated. HepCore displayed the following sequence: TPPAYRPPNAPIL. All peptides were dissolved in sterile PBS and stored at −80 °C.

Generation of bone marrow–derived DCs

Murine bone marrow was isolated from tibias and femurs of female C57/Bl6 mice. Bones were collected and flushed with RPMI-1640 and the suspension was filtered through a 70-μm cell strainer (BD Falcon). Cells were resuspended in CM supplemented with 1% 2-mercaptoethanol and adhered for 24 h in 6-well plates. Next day, loosely adherent and non-adherent cells were collected and adjusted to 106 cells/ml in CM and supplemented with 1% 2-mercaptoethanol, 10 ng/ml murine GM-CSF and 20 ng/ml IL-4 (both Peprotech). Wells were replenished with fresh medium and cytokines on day 3 of culturing and at day 6 loosely adherent, and non-adherent cells were collected. For tumour challenge experiments, FCS was replaced with 1.5% autologous mouse serum during DC culturing and DC contact with FCS or BSA was avoided.

Immunization and tumour challenge

Day 6 immature DCs were adjusted to 1.5*106 cells/ml in CM supplemented with 1% 2-mercaptoethanol and cytokines as above and cells were maturated by incubation with 0.5 μg/ml lipopolysaccharide (LPS) for 24 h (Sigma Aldrich, Brøndby, Denmark). Peptides were added in a final concentration of 10 μm during the last 5 h of incubation. After this, DCs were harvested, washed in PBS and injected s.c. at the base of the tail as 106 cells pr mouse in PBS + 1%BSA. Vaccines were provided with 7 days interval and lymphoid organs were harvested 7 days after the last vaccination. In the tumour challenge experiments, these DCs were generated in autologous mouse serum. In separate experiments, DCs were loaded with 200 ng alpha-galactosylceramide overnight followed by addition of the MHC class–I-binding peptides for an hour and then washed three times before i.v. injection.

In separate experiments, immunization was performed in Incomplete Freund's Adjuvant (IFA). Fifty microgram MHC class-I peptides (TAYQSFEQV and SIINFEKL) and 50 μg HepCore were injected subcutaneously in the groin region of female C57Bl/6 mice in 200 μl emulsion of equal volumes of PBS and IFA (Sigma Aldrich). Two booster immunizations were given with 7–10 days interval in similar emulsion.

In tumour challenge experiments, DCs were generated in autologous serum and tumour cells (2*105 B16 cells) were injected s.c. in the right flank as indicated. Tumour growth was measured daily by the use of a calliper, and the total tumour volume was calculated as V = (a*b)2/2 with a being the largest tumour diameter and b the smallest tumour diameter. Mice were euthanized when the tumour volume exceeded 500 mm3.


For all analysis, a gate was set at large granular and viable cells based on forward- and side-scatter. 10.000 events in the gate were analysed.

Surface marker staining: After washing, harvested cells were subjected to CD16/CD32 blocking using purified antibody (BD) for 15 min on ice, followed by washing. Surface staining antibodies were applied and samples were incubated for 30 min on ice, in the dark. Samples were washed four times and finally resuspended in PBS prior to data acquisition. Antibodies used: H-2Kb-FITC, H-2Db-PE, CD90.2-FITC, CD44-PE, CD11c-FITC, CD80-PE, CD86-PE, MHC II-PE and CD40-APC (all from BD Bioscience, Albertslund, Denmark).

Intracellular Bcl-xL staining: Adherent cells were trypsinized before the staining. Following washing in PBS, cells were fixated and permeabilized in 100 μl CytoFix/CytoPerm (BD) for 30 min on ice in the dark. Samples were washed twice in Perm/Wash (BD) and stained with anti-Bcl-xL-PE antibody (AbCam, Cambridge, UK). Cells were incubated for 30 min on ice and washed prior to analysis. Recommended isotype control antibodies (BD) were applied in all experiments.

In vitro peptide restimulation, intracellular cytokine staining and cytotoxicity

Draining lymph nodes (LNs) and spleens were recovered and immediately transferred to RPMI 1640. Cells were isolated by squeezing LNs and spleens through a 70-μm nylon cell strainer using the syringe from a cannula. Next, cells were resuspended in RPMI + 10% FCS supplemented with 1% 2-mercaptoethanol (2-ME) and plated in the wells of a 96-well round-bottomed plate. Positive control wells were further supplemented with phorbol 12-myristate 13-acetate (PMA from Sigma Aldrich) (50 ng/ml), ionomycin (750 ng/ml from Sigma Aldrich), IL-2 (50 U/ml) and Golgi-stop (BD) (0.67 μl/ml) which provided an unspecific polyclonal stimulation of cells. All other samples received IL-2 (50 U/ml) and Golgi-stop (0.67 μl/ml). Reactivity against specific peptides was tested by addition of 10 μg/ml peptide. Samples were incubated at 37 °C, 5% CO2 for 5 h to stimulate cytokine production. Following incubation, surface staining was performed prior to permeabilization and intracellular staining according to protocols as described previously.

Antibodies used: CD4-FITC, CD8-APC, CD8-FITC, IFN-γ-PE, IFN-γ-PE-Cy7, IL-4-PE, IL-17A-PE and IL-10-PE (all from BD). In all experiments, data acquisition was performed on a FACSCalibur flow cytometer (BD), and analysis was performed using CellQuest software (BD). To test cytoxicity of spleen cells from immunized cells, spleen cells were expanded for 8–10 days with peptide and IL-2. The relevant cancer cell lines were labelled with radioactive chromium and used as target cells in a standard 4 h chromium release assay where peptide expanded spleen cells were used as effector cells.

Western blotting

Mitochondrial fractionation and lysis: Harvested cells were washed in PBS prior to resuspension in 50 μl 0.33 m sucrose buffer (Sucrose, EDTA and 15 mm tris-HCl). The suspension underwent four freeze/thaw cycles (−80 °C/37 °C) prior to centrifugation at 800 g, 4 °C for 10 min. The supernatant was transferred to new tubes and centrifuged at 8200 g, 4 °C for 10 min. The remaining pellet was resuspended in 30 μl RIPA-buffer and incubated at room temperature for 20 min, including gentle mixing by a vortex machine twice during incubation. The lysate was centrifuged at 8200 g, 4 °C for 10 min.

Western Blot analysis: Cell lysate was mixed with 6 X sample loading buffer (50% glycerol, 0.1 g/ml SDS, 25% 0.5 m Tris pH 6.8, 0.05 mg/ml bromphenol blue). Samples were heated for 5 min followed by loading onto 4–12% Bis-Tris gels (Invitrogen, Nærum, Denmark). Electrophoresis was performed in 1X MES running buffer (Invitrogen) for 1 h at 200 V. Blotting was performed at 30 V for 1 h onto nitrocellulose membranes (Invitrogen) in transfer buffer (Invitrogen). Nitrocellulose membranes were blocked for 1 h at room temperature in PBS containing 3% dried non-fat milk and then probed with primary antibodies (Bcl-xL-specific antibody, H-62 rabbit polyclonal antibody) was from Santa Cruz Biotechnology Dallas, TX, USA, XIAP [polyclonal rabbit antibody, survivin (71G4B7 rabbit mAb) and bcl-2 (D17C4 Rabbit mAb)-specific antibodies were from Cell Signalling Technology, Boston, MA, USA] dissolved according to the manufacturer's instructions in 1% tris-buffered saline (TBS) containing 5% bovine serum albumin (BSA) (Sigma Aldrich) and 0.1% Tween20 (BIO-RAD Laboratories, Inc., Copenhagen, Denmark) over night at 4 °C. Secondary horse radish peroxidase (HRP)-conjugated antibodies was dissolved in 1% TBS, 5% BSA and 0.1% Tween20 and membranes were incubated at room temperature for 1 h. Secondary HRP-conjugated antibodies were rabbit anti-goat and goat anti-rabbit antibodies (both DAKO). The membrane was washed three times in washing buffer, twice in PBS and visualized using an enhanced chemiluminescence detection system (ECL Plus Western Blotting Detection system, GE Healthcare) according to the manufacturer's instructions. Detection of HRP-generated precipitates was performed on Typhoon Scanner 9410 with Typhoon Scanner software (both from Amersham Biosciences, Piscataway, NJ, USA).


Student's t-test was used for the comparison of the mean, and data were considered significantly different at P < 0.05.


Bcl-xL expression in murine cancer cell lines

The anti-apoptotic protein Bcl-xL may be a potent tumour antigen, because overexpression might facilitate overt cancer outgrowth in a hostile tumour microenvironment. To identify model tumour cell lines for Bcl-xL-based cancer vaccination studies, we investigated Bcl-xL expression in various tumour cell lines. Using western blotting (Fig. 1A), we detected expression of Bcl-xL in B16 melanoma cells and in EL4 thymoma cells. Western blotting was performed on mitochondrial fractionated cell lysates from B16 and EL4 cells. As shown, Bcl-xL was detected both as monomeric protein (26 kDa, arrow) and in complexes of higher molecular weight (Fig. 1A, arrowheads). Bcl-xL expression in these cell lines was confirmed with flowcytometry (Fig. 1B). Bcl-xL was also expressed in mouse splenocytes (data not shown). Furthermore, we performed a screening for the expression of other anti-apoptotic proteins and detected expression of XIAP, survivin and Bcl-2 in B16 cells, while EL4 cells expressed survivin and Bcl-2 (Fig. 1C).

Figure 1.

Expression of inhibitors of apoptosis and surface MHC class-I expression in B16 and EL4 cells. (A) Western blot of mitochondrial fractionated cell lysates showing expression of Bcl-xL in B16 and EL4 cells. Bcl-xL was detected as monomeric protein (26 kDa, arrow) and in complexes of higher molecular weight (arrowheads). Lysates were obtained from 5-20*106 (B16) or 5-25*106 cells (EL4) and lysates from increasing cell numbers were added from left to right on the gel. (B) Intracellular flow cytometry showing Bcl-xL expression (black) compared with isotype control staining (black line) in B16 and EL4 cells. Numbers indicate the percentage of positive cells. (C) Western blots showing expression of multiple inhibitors of apoptosis in B16 and EL4 cells. (D) Flow cytometric analysis of MHC class-I expression in B16 and EL4 cells. Numbers indicate the percentage of cells in indicated quadrants.

Overt cancer outgrowth is further facilitated by the downregulation of tumour cell MHC expression. We observed low levels of the MHC class-I alleles H-2Kb and H-2Db in B16 cells (Fig. 1D). MHC class-I expression could, however, be induced in B16 cells after in vitro IFN-γ treatment (Fig. 1D). In contrast to B16 cells, 98% of EL4 cells stained positive for H-2Db, while 19% of all EL4 cells were H-2Kb+H-2Db double positive even in the absence of IFN-γ treatment.

In silico prediction of Bcl-xL-derived H-2b binders

To predict putative Bcl-xL-derived MHC class–I-binding epitopes in the murine setting, the insert sequence discriminating anti-apoptotic Bcl-xL from pro-apoptotic Bcl-xS (AA 126-188) including a 10 AA N- and C-terminal overlap with the common Bcl-x sequence was scrutinized for putative H-2b binders using the in silico algorithm NetMHC3.2. Based on NetMHC predicted affinities (Table 1), the putative H-2Kb binders TAYQSFEQV (AA118-126) and AFFSFGGAL (AA143-151) were selected as candidate epitopes for the induction of specific CD8+ T-cell responses.

Table 1. List of epitopes used in vaccination studies and predicted affinity of binding to the relevant MHC class-I allele. Predicted binding affinities were obtained using the in silico predictor algorithm NetMHC3.2
ProteinEpitope sequenceMHC-I alleleAffinity (nm)

DC-based vaccination is superior to incomplete Freund's adjuvant for the induction of IFN-γ-producing CD8+ T cells

We compared vaccination with peptide-pulsed bone marrow–derived DCs to vaccination with peptides delivered in an emulsion of PBS and IFA. Here, in vitro generated CD11c+ DCs displayed high surface expression of the DC activation markers CD80, CD86, MHC class-II molecules and CD40 upon 24 h of LPS maturation (Fig. 2A). All of these activation markers are functionally relevant in the induction of TH1-polarized T-cell immunity. Furthermore, DCs maturated with LPS displayed a favourable cytokine secretion profile, as IL-12p70 was secreted in large amounts after LPS treatment. Importantly, IL-10 production by DCs decreased during the gradual in vitro maturation with GM-CSF and IL-4, with IL-10 production being undetectable in LPS-maturated DCs (data not shown).

Figure 2.

DC-based vaccination is superior to IFA-based vaccination in inducing CD8+ T-cell responses. (A) Analysis of maturation marker expression on immature (iDC) or LPS-maturated DCs (mDCs). Numbers indicate the percentage of CD11c+ cells expressing the indicated surface marker. (B) Intracellular flow cytometry of CD8+ T cells recovered from mice having received three subcutaneous vaccinations with SIINFEKL-pulsed DCs or SIINFEKL in IFA as indicated. T cells were restimulated with SIINFEKL 5 h prior to analysis. Numbers indicate the percentage of IFN-γ+CD8+ T cells. LN = lymph nodes. (C) Quantification of the percentage of IFN-γ+CD8+ T cells obtained from the indicated conditions and compartments. (n = 3). *P > 0.05.

DC-based vaccination was compared with IFA-based vaccination using the ovalbumin-derived H-2Kb-binding peptide SIINFEKL (Table 1). Peptide restimulation of freshly isolated T cells from mice having received three vaccinations revealed that the DC-based vaccine induced significantly larger numbers of SIINFEKL-reactive CD8+ T cells producing IFN-γ (Fig. 2B). A significantly larger percentage of SIINFEKL-reactive CD8+ T cells were found in both draining lymph nodes and spleens of DC-vaccinated mice compared with IFA-vaccinated animals. Interestingly, the percentage of SIINFEKL-specific CD8+ T cells was significantly larger in spleens than in draining lymph nodes (Fig. 2C). This suggests that primed SIINFEKL-specific T cells had already left the draining lymph node and entered the systemic circulation at the time of organ harvest.

Induction of Bcl-xL-specific IFN-γ-producing CD8+ T cells

Next, we applied the DC-based vaccination strategy to epitopes derived from the mouse Bcl-xL protein (Table 1). Three sequential subcutaneous vaccinations with TAYQSFEQV- or AFFSFGGAL-pulsed DCs induced large numbers of CD8+ T cells producing IFN-γ upon in vitro restimulation (Fig. 3). Interestingly, a single vaccination with TAYQSFEQV and AFFSFGGAL double-pulsed DCs generated large numbers of antigen-specific CD8+ T cells producing IFN-γ with percentages being as high as 4.84% of CD8+ T cells in one animal (TAYQSFEQV reactivity) (Fig. 4). There was a trend that the presence of the MHC class-II helper peptide HepCore on pulsed DCs increased both the level of IFN-γ and the percentage of IFN-γ secreting TAYQSFEQV and AFFSFGGAL reactive CD8+ T cells (Fig. 4), although the cohorts of mice were two small to detect significant differences. In contrast, transplantation with, for example, B16 tumours did not in itself induce induction of Bcl-xL-specific CD8+ T cells (data not shown) demonstrating the need for immunization.

Figure 3.

Induction of CD8+ T cells responses against Bcl-xL-derived epitopes TAYQSFEQV and AFFSFGGAL. (A) Flow cytometric analysis of IFN-γ production in CD8+ T cells from mice vaccinated three times with TAYQSFEQV-pulsed DCs. Isolated splenocytes were restimulated for 5 h with TAYQSFEQV before analysis (n = 4). (B) Flow cytometric analysis of IFN-γ production in CD8+ T cells from mice vaccinated three times with AFFSFGGAL-pulsed DCs or unpulsed DC (vehicle). Isolated splenocytes were restimulated for 5 h with AFFSFGGAL before analysis (n = 3).

Figure 4.

A single vaccination with TAYQSFEQV, AFFSFGGAL double-pulsed DCs induces high numbers of IFN-γ-producing CD8+ T cells. (A) Flow cytometric analysis of IFN-γ production in CD8+ T cells after restimulation with either TAYQSFEQV or AFFSFGGAL. Mice received a single vaccination with TAYQSFEQV, AFFSFGGAL double-pulsed DCs with or without HepCore pulse. T cells were recovered from lymph nodes or spleens as indicated 1 week after vaccination. Numbers indicate the percentage of IFN-γ+CD8+ T cells. (B–E) Quantification of the percentage of lymph node (B) or spleen-recovered (D) CD8+ T cells producing IFN-γ after restimulation with TAYQSFEQV or AFFSGGAL and the MFI-values of IFN-γ+ CD8 T cells from lymph nodes (C) or spleens (E) from vaccinated mice (n = 3).

Previous work has demonstrated that intravenous immunization with antigen-loaded DC cotreated with the NKT cell ligand alpha-galactosylceramide can induce potent CD8+ T-cell responses [17], with greater number of circulating antigen-specific CD8+ T cells in comparison with subcutaneous immunization with LPS-matured antigen-loaded DC (data not shown). Thus, we tested this approach for improved induction of Bcl-xL-specific CTLs. However, while peptide-loaded DCs cotreated with the NKT cell ligand alpha-galactosylceramide-induced TAYQSFEQV- and AFFSFGGAL-specific CD8+ T-cell responses, the response was similar in size to that induced by peptide-loaded DCs (data not shown). This finding possibly reflects the relatively low MHC class-I-binding affinity for the selected peptides and suggests that increased MHC binding affinity by P2 anchor modification of the peptides is necessary to augment the capacity of the peptides to induce CD8+ T-cell responses.

Functional effect of Bcl-xL-specific CD8+ T cells

Splenocytes isolated from mice having received vaccination with DCs pulsed with TAYQSFEQV and AFFSFGGAL displayed significantly greater cytotoxicity as compared to splenocytes from naïve mice, in particular against the Bcl-xL-expressing cell line B16. However, cytotoxicity against EL4 was not observed (Fig. 5A). Thus, despite Bcl-xL and high MHC class expression other factors, for example, expression profile of anti-apoptotic proteins, may prevent CTL-mediated cytotoxicity. Also, the observed cytotoxicity against B16 could in part be NK mediated.

Figure 5.

Bcl-xL-specific T cells are cytotoxic against tumour cells pulsed with TAYQSFEQV and AFFSFGGAL. (A) 51Cr-release from B16 or EL4 cells after 4 h of coculture with splenocytes (pre-expanded with peptide and IL-2) recovered from mice having received vaccination with TAYQSFEQV, AFFSFGGAL double-pulsed DCs or from naïve controls. (B) Flow cytometric analysis of cytokine production from CD4+ and CD8+ T cells isolated from subcutaneously transplanted B16 tumours. Cytokine production was analysed after 5 h of polyclonal stimulation with PMA and ionomycin. Numbers indicate the percentage of T cells producing the indicated cytokine.

We also investigated the function of spontaneous Bcl-xL T-cell responses. Initial B16 challenge studies in naïve mice revealed prominent infiltration of both CD4+ and CD8+ T cells in subcutaneously transplanted tumours which produced high amount of IFN-γ without IL-10 or IL-17 secretion (Fig. 5B). However, a CD8+ T-cell response against TAYQSFEQV or AFFSFGGAL could not be detected in the tumour microenvironment of naïve B16 challenged mice (data not shown), suggesting that these novel epitopes were not yet utilized by tumour infiltrating T cells. Thus, vaccine-induced CTLs against these novel epitopes are like to increase existing spontaneous anti-tumour immune responses.

However, prophylactic DC vaccination before transplantation with this particular cell line was unsuccessful, also when DCs were cotreated with the NKT cell ligand alpha-galactosylceramide (data not shown) which have previously been described as a promising approach [17]. Thus, further vaccine development and definition of optimal cotreatment with additional immunotherapy is required. In particular, the latter is very often required for cancer vaccines to be effective [18, 19].


Establishment of mouse tumour vaccination models is pivotal for preclinical development of cancer vaccines and immunotherapy. Here, we identified that the tumour antigen Bcl-xL was endogenously expressed by murine tumour cell lines EL4 and B16 and that Bcl-xL-specific CD8+ T cells could be activated indicating that tolerance to this self-antigen could be broken by DC-based vaccination. Thus, these epitopes are likely to be used in future preclinical studies of cancer vaccination when endogenous tumour antigens are required for well-characterized cell lines.

Cancer vaccines were brought an important step forward with the FDA approval of Provenge (Sipuleucel-T) against metastatic prostate cancer [4]. Also, immunotherapy with Interleukin-2, IFN-α [20] and anti-CLTA4 (Ipilimumab) [1] as well as adoptive T-cell transfer [21] has shown promising results. In the future, it is expected that a large number of combinations with such immunotherapy and with established chemotherapy, for example, cyclophosphamide [22] or novel small drugs, for example, Vemurafenib [23] will be tested and increase response rates in various cancer forms. However, for design of such human clinical studies, input from preclinical cancer vaccine and immunotherapy models is pivotal. Thus, although we did not observe prophylactic effect with this vaccine alone, it is likely that identified Bcl-xL-specific vaccine response will have a role in preclinical studies of combined cancer immunotherapy.

Spontaneous Bcl-xL-reactive CD8+ T cells were recently identified in patients with breast cancer, malignant melanoma and pancreatic cancer, whereas no reactivity was observed in healthy controls [15]. Such CD8+ T cells were demonstrated to kill both peptide-pulsed antigen-presenting cells and Bcl-xL-expressing human tumour cell lines [16]. Of importance, these studies showed that Bcl-xL expressed on B and T cells were not targeted, suggesting that specific CD8+ T cells selectively attacked the Bcl-xL epitope on tumour cells, probably due to alternative protective anti-apoptotic signals in leucocytes. Recent findings suggest that melanoma cells may dedifferentiate in an inflammatory environment and thus escape recognition from T cells specific for classical melanoma differentiation antigens [24]. This highlights that Bcl-xL, as an anti-apoptotic protein, might be a better alternative as tumour antigen candidate, as it belongs to a family of tumour antigens that cannot be downregulated during tumour progression. We did not find spontaneous Bcl-xL-specific responses in tumour-bearing mice, probably reflecting the limited amount of tissue damage and tumour spreading from these transplanted tumours as compared to human tumours.

Using a DC-based vaccine in mice, we observed that IFN-γ-producing CD8+ T cells could be induced against murine Bcl-xL. As such, we have provided an important preclinical cancer vaccine model taking advantage of an endogenous tumour antigen. Despite, the induction of Bcl-xL-specific IFN-γ-secreting CD8+ T cells, the Bcl-xL-specific in vitro cytotoxicity was low and NK-dependent mechanism as part of this response could not be ruled out, because the MHC class-I low-expressing tumours were more proned to killing. Also, the use of Bcl-xL-silenced cell lines could have provided additional proof for the Bcl-xL specificity.

The DC-based vaccine alone was not able to induce regression of EL4 or B16 Bcl-xL-expressing tumour growth in mice, even when tumour cells where pretreated with IFN-γ before transplantation (data not shown). In line with this, previous studies on DC vaccination with the MuLV gp70 envelope protein-derived peptides AH1 and p320-333 (another endogenous antigen) against the murine colon cancer cell line CT26 also failed to eradicate tumours when used alone but demonstrated synergistic effect when combined with an antibody against VEGF receptor 2 which may block the DC suppression from tumour-derived VEGF [25]. Similar, combination with anti-CTLA4 treatment seems to enhance vaccine efficacy dramatically [18, 19]. Also, several other tumour immune suppression mechanism exist in vivo such as, for example, recruitment of regulatory T cells and myeloid derived suppressor cells to the tumour site [26] which may explain the discrepancy between our in vitro cytotoxicity findings and the lack of tumour regression in vivo. However, it is likely that increased efficacy can be obtained both in relation to in vitro cytotoxicity and in vivo effect on tumour challenge by combination with additional cancer treatments, and the model may therefore still be an interesting platform for studies that test the synergy between cancer vaccination and other cancer treatments such as other immunotherapy and chemotherapy. As an example of this, anthracylines sensitizes tumour cells to increased uptake by dendritic cells, and cyclophosphamide have in some studies been shown to clear regulatory T cells thus potentiating the effects of, for example, cancer vaccines [27]. For the translation of such findings into clinical studies, it is of importance with good preclinical in vivo tumour models. Similar, improvements in adoptive T-cell transfer, which have recently been shown successful against malignant melanoma in humans [21], may also benefit from potent mouse models using endogenous tumour antigens such as Bcl-xL. Alternatively, the vaccine in itself may potentially be improved by P2 anchor modification of the peptides to increase MHC-binding affinity [28, 29].

In conclusion, we identified the H-2Kb-binding peptides TAYQSFEQV and AFFSFGGAL from murine Bcl-xL. These peptides induced Bcl-xL-specific IFN-γ-producing CD8+ T cells when used in a DC-based vaccination. These peptides have relevance in preclinical mouse models of cancer immunotherapy and enables vaccination studies of an endogenous tumour antigen. Preclinical studies using such endogenous tumour antigens may also identify optimal combinations of cancer immunotherapy which can be translated to a clinical situation.


This project was supported by University of Copenhagen and a grant from ‘Fabrikant Einar Willumsens Mindelegat’.

Conflict of interest

The authors declare no conflict of interest.