Abstract
Cancer vaccines aim to induce CTL responses against tumors. Challenges for vaccine design are targeting Ag to dendritic cells (DCs) in vivo, facilitating cross-presentation, and conditioning the microenvironment for Th1 type immune responses. In this study, we report that ISCOM vaccines, which consist of ISCOMATRIX adjuvant and protein Ag, meet these challenges. Subcutaneous injection of an ISCOM vaccine in mice led to a substantial influx and activation of innate and adaptive immune effector cells in vaccine site-draining lymph nodes (VDLNs) as well as IFN-γ production by NK and NKT cells. Moreover, an ISCOM vaccine containing the model Ag OVA (OVA/ISCOM vaccine) was efficiently taken up by CD8α+ DCs in VDLNs and induced their maturation and IL-12 production. Adoptive transfer of transgenic OT-I T cells revealed highly efficient cross-presentation of the OVA/ISCOM vaccine in vivo, whereas cross-presentation of soluble OVA was poor even at a 100-fold higher concentration. Cross-presenting activity was restricted to CD8α+ DCs in VDLNs, whereas Langerin+ DCs and CD8α− DCs were dispensable. Remarkably, compared with other adjuvant systems, the OVA/ISCOM vaccine induced a high frequency of OVA-specific CTLs capable of tumor cell killing in different tumor models. Thus, ISCOM vaccines combine potent immune activation with Ag delivery to CD8α+ DCs in vivo for efficient induction of CTL responses.
Tumor vaccines seek to induce a CTL response against tumors. To achieve efficient tumor cell killing, different strategies have been evaluated for inducing both CD4+ Th cell and CD8+ T cell responses against tumor Ags. Attention has focused on exploiting dendritic cells (DCs) as professional APCs capable of presenting exogenous Ag not only on MHC class II, but also on MHC class I, a process termed cross-presentation. Because of feasibility issues, most vaccination protocols have relied on in vitro-generated DCs loaded with tumor Ag as an individualized tumor vaccine (1, 2). As DC activation has been recognized to be of critical importance for the induction of productive immune responses, DC maturation was included in vaccination protocols (3, 4). However, the production of DC vaccines is labor- and cost-intensive, as cytapheresis may be required and vaccines have to be manufactured individually for each patient. A promising strategy circumventing the need of ex vivo DC generation and manipulation is targeting vaccines directly to DCs in vivo (5, 6). Moreover, the development of cell-free vaccines for off-the-shelf use will make vaccines accessible to more patients and reduce manufacturing costs (7). Challenges for the design of effective cancer vaccines are DC-specific Ag targeting, facilitating MHC class I epitope processing, and identifying adjuvants that activate DCs in vivo for generating and activating effector T cells as well as innate effector cells.
An important parameter for vaccine design is the choice of tumor Ag and its physical properties (i.e., peptide, protein, DNA or RNA). Most clinical trials have evaluated synthetic MHC class I- and/or MHC class II-restricted peptides derived from tumor Ags. A drawback of using preformed peptides is restricting treatment to patients with a limited number of MHC haplotypes. This limitation can be circumvented by using full-length proteins, which 1) contain multiple MHC class I and MHC class II epitopes, 2) can be used without knowledge of the patient’s MHC haplotype, and 3) are presented for prolonged periods on MHC molecules as compared with peptides (8). A shortcoming may be that DCs cross-present soluble protein inefficiently on MHC class I. However, this limitation can be circumvented by using Ab–Ag conjugates that target DC surface receptors involved in Ag uptake, such as Fc receptors (8–11), members of the C-type lectin receptor family including CD205, the mannose receptor, and DC-specific intracellular adhesion molecule 3-grabbing nonintegrin (12–14).
Alternatively, Ag can be formulated with chemically defined delivery systems, such as ISCOMATRIX adjuvant, which is derived from the immunostimulating complex (ISCOM) first described by Morein et al. (15). They observed enhanced immune responses to protein formulated with a mixture of saponin, phospholipids, and cholesterol that forms cage-like structures (15). ISCOMATRIX vaccines that contain only a purified fraction of quillaia saponin were shown to be safe, well tolerated, and highly immunogenic in humans, generating Ab and T cell responses (16). An ISCOM vaccine has the Ag incorporated into the structure during manufacture, whereas an ISCOMATRIX vaccine is made by mixing Ag with preformed ISCOMATRIX adjuvant. Both entities are otherwise identical. Interestingly, ISCOM vaccines facilitate cross-presentation by DCs via translocation of Ag from endosomes into the cytosol (8, 17). Moreover, ISCOMATRIX adjuvant-based tumor vaccines containing full-length tumor Ag-induced humoral and cellular immune responses in mouse models (18, 19) as well as in cancer patients (20). Taken together, these studies demonstrate the high clinical potential of ISCOMATRIX adjuvant in tumor vaccines.
Little is known about the immunological effects mediated by ISCOM and ISCOMATRIX vaccines on DCs and other leukocyte populations in vivo. Understanding the mechanisms of action underlying these vaccines will be important for optimal use in the clinic and further optimization. In this study, we investigated the effects of ISCOMATRIX adjuvant and an ISCOM vaccine containing the model Ag OVA on cytokine milieu and on phenotype as well as function of innate and adaptive immune effector cells in vaccine site-draining lymph nodes (VDLNs). We further studied Ag uptake and presentation by distinct DC populations in VDLNs and analyzed the induction of cellular immune responses in wild-type mice and mice in which distinct DC populations were depleted.
Materials and Methods
Cell culture media and reagents
Primary cells were cultivated in RPMI 1640 media supplemented with heat-inactivated 10% FCS (Life Technologies BRL, Paisley, U.K.), 2 mM l-glutamine, penicillin (100 U/l), streptomycin (0.1 mg/ml), 100 μM nonessential amino acids, 1 mM sodium pyruvate (all PAA, Linz, Austria), and 50 μM 2-ME. The OVA/ISCOM vaccine was generated by associating palmitified OVA (Sigma-Aldrich, St. Louis, MO), which in some experiments was labeled with the fluorochrome Alexa 488, into the ISCOM structure by formulation in the presence of ISCOPREP saponin, phospholipid, and cholesterol, as previously described (17). ISCOMATRIX adjuvant was made in the same way without the OVA. The H2-Kb–restricted peptide OVA257–264 (SIINFEKL) was purchased from Jerini Peptide Technologies (Berlin, Germany). CpG-containing oligodeoxynucleotide 1826 (CpG-ODN1826; InvivoGen, San Diego, CA) and Pam3-Cys (InvivoGen) were used at 5 μg, LPS (LPS 0111:B4; Chondrex, Redmond, WA) at 1 μg, and aluminum hydroxide (Aluhydrogel; Superfos Biosector, Frederikssund, Denmark) at 100 μg per injection. IFA (Bacto/Difco, Mount Pritchard, Australia) was used at a 50:50 dilution of the stock solution.
Mice and vaccinations
Female C57BL/6 mice, 6–8 wk old, were purchased from Harlan Winkelmann (Borchen, Germany). Langerin-diphtheria toxin receptor (DTR)/enhanced GFP knock-in mice on C57BL/6 background expressing a DTR and enhanced GFP under the Langerin promoter were provided by Dr. B.E. Clausen (21). For depletion of Langerin+ DCs we administered 500 ng DT (Sigma-Aldrich) 1 d prior to immunization. A week later DT injection was repeated to keep Langerin+ DCs absent throughout the experiment. For depletion of CD11c+ DCs, transgenic mice expressing the DTR fused with GFP under control of the CD11c promoter (CD11c-DTR mice) were used (22). Irradiated C57BL/6 mice were reconstituted with bone marrow from CD11c-DTR transgenic mice, rendering DTR expressing CD11c+ DCs from hematopoietic origin sensitive to DT treatment. Reconstituted mice were administered three doses of DT every 3 d from day 0. An 80% depletion of CD11c+ DCs was confirmed before the first immunization. Transgenic OT-1 mice were provided by Prof. T. Brocker (Department of Immunology, University of Munich, Munich, Germany). Animal experiments were approved by the local regulatory agencies.
For vaccination, mice were injected s.c. into the lower hind leg with 50–100 μl PBS containing either OVA (30 μg) with or without adjuvants, ISCOMATRIX adjuvant (5 μg), or OVA/ISCOM vaccine (containing 0.3 μg OVA/5 μg ISCOM). Vaccination was repeated on day 7 (prime-boost regimen).
LN preparation
Mice were anesthetized with 1-chlor-2,2,2-trifluorethyl-difluormethyl ether (Forene; Abbott, Wiesbaden, Germany) and killed by cervical dislocation. The VDLNs were removed and processed into single-cell suspensions by passing through a 70-μm cell strainer. Adjusted cell numbers were processed for phenotypic and functional analysis.
mAbs and flow cytometry
For flow cytometry, cell suspensions were incubated for 20 min at 4°C with the following Abs: CD3ε (clone 145-2C11), CD4 (clone GK1.5), CD8α (clone 53-6.7), CD11b (clone M1-70), CD11c (clone HL3), IL-12p40 (clone C15.6), CD19 (clone 1D3), CD86 (clone GL1), NK1.1 (clone PK136; all from BD Biosciences, San Jose, CA); CD69 (clone H1.2F3) and IFN-γ (clone XMG1.2; both from Caltag Laboratories, Carlsbad, CA); and Ly6G (clone RB6-8C5; eBioscience, San Diego, CA) and MHC class II (clone NIMR-4; SouthernBiotech, Birmingham, AL). Samples were acquired on a FACSCalibur (BD Biosciences, Heidelberg, Germany) and analyzed using FlowJo version 7.2.1 (Tree Star, Ashland, OR).
Intracellular cytokine staining and pentamer staining
To assess cytokine expression by DCs, NK cells, and NKT cells, cell suspensions of VDLNs were incubated with 1 μg/ml brefeldin A (Fluka, Munich, Germany) for 4 h at 37°C and subsequently stained for surface markers, fixed with 2% paraformaldehyde, incubated with mAb against IL-12p40 or IFN-γ in 0.5% saponin in PBS, and analyzed by flow cytometry. To measure OVA-specific CTL response, PBMCs of 100–200 μl peripheral blood were stimulated with SIINFEKL peptide (1 μg/ml) for 1 h at 37°C before addition of 1 μg/ml brefeldin A. Three hours later IFN-γ intracellular cytokine staining (ICS) was performed as described above. Alternatively, the number of OVA-specific CTLs was determined using SIINFEKL/H-2Kb pentamer (ProImmune, Oxford, U.K.) together with mAbs against CD8α, CD19, and NK1.1. Unspecific binding of pentamer by dead cells and CD19+ B cells/NK1.1+ NK cells was excluded by gating on 7-aminoactinomycin D−CD19−NK1.1− cells before analysis as recommended by the manufacturer’s protocol.
Cytokine measurement
VDLNs were shock frozen in liquid nitrogen, processed using mortar and pestle, and lysed with 50 μl protein lysis buffer (Bio-Rad, Munich, Germany). Samples were vortexed for 30 s, centrifuged for 15 min at 12,000 × g and 4°C, and supernatant was used for cytokine analysis. Samples were adjusted to 2 mg protein/ml as determined by Bradford assay (Bio-Rad). IL-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p40, IL-12p70, TNF-α, IFN-γ, and GM-CSF concentrations were determined by a Bio-Rad multiplex suspension array according to the manufacturer’s instructions. For Western blot analysis, adjusted protein samples were separated by SDS-PAGE. IL-1β was detected using a primary goat anti-mouse IL-1β Ab (R&D Systems) followed by a secondary HRP-coupled anti-goat Ab (Santa Cruz Biotechnology, Heidelberg, Germany). β-actin served as a loading control. IL-1β in supernatants of bone marrow-derived DCs was measured by ELISA (BD Biosciences). DCs (106 cells/ml) were treated with 200 ng/ml LPS for 2 h before addition of 10 μM caspase-1 inhibitor z-YVAD-fmk (Calbiochem, Darmstadt, Germany) for 1 h followed by 6 h incubation with 5 μg/ml ISCOMATRIX adjuvant.
In vitro cross-presentation assay
Total DCs were enriched by Percoll density gradient centrifugation (1.077 g/cm3; Biochrom) after mechanical and enzymatic processing of spleens. Untouched CD8α+ DCs were obtained by combining the Ab mixture of the mouse DC isolation kit (Invitrogen) with biotinylated anti-CD4 (clone RM4-5) and anti-B220 (clone RA3-6B2; both from BioLegend). Finally, cells were enriched by magnetic separation using streptavidin-coupled beads (Invitrogen). Purity was usually >90% as determined by flow cytometry. CD8α+ DCs were incubated with the indicated amounts of OVA or OVA/ISCOM vaccine for 90 min and washed four times in serum-free media. Finally, 2.5 × 104 DCs were used as stimulators for 5 × 104 enriched CFSE-labeled OT-I T cells. Proliferation from viable CD8+ T cells was determined after 3 d (23).
In vivo T cell proliferation assay
Splenocytes from OT-1 mice were suspended at 5 × 107 cells/ml in PBS/0.1% BSA containing 10 μM CFSE (Invitrogen/Molecular Probes, Eugene, OR) for 10 min at 37°C. Cells were washed twice with cold RPMI 1640/10% FCS followed by two washes in PBS. CFSE-labeled OT-1 cells (2 × 106) in 200 μl PBS were injected into the tail vein of mice that were vaccinated once with OVA, OVA/ISCOM vaccine, or PBS on the same day. Two days later, peripheral blood was collected and CFSE dilution of proliferating OT-1 CD8+ T cells was analyzed by flow cytometry.
Ex vivo cross-presentation by DC subpopulations
Twenty-four hours after s.c. immunization, VDLNs were processed into a single-cell suspension with a collagenase/DNAase solution. DCs were enriched by Nycodenz density gradient centrifugation (1.082 g/cm3; Nycomed Pharma, Oslo, Norway), followed by depletion of non-DCs and plasmacytoid DCs with an Ab mixture containing anti-CD3 (clone KT3), anti–Thy-1 (cloneT24/31.7), anti-CD19 (clone ID6), anti-Gr1 (clone RB6-8C5), and Ter-119 and magnetic beads coated with anti-rat IgG (Dynal beads M450; Invitrogen). Enriched DCs were stained with mouse anti–CD11c-allophycocyanin, anti–CD8-FITC, and anti–CD205-PE (BD Biosciences) and FACS sorted into three DC subpopulations (FACSAria; BD Biosciences). A purity of >93% was obtained routinely. DCs were cultured with CFSE-labeled OT-I cells at 10,000:50,000 DC/OT-I cell ratio, and proliferation of OT-I cells was measured 48 h later by flow cytometry.
In vivo cytotoxicity assay
Mice were injected i.v. with a mixture of splenocytes differentially labeled with CFSE (2, 20, or 200 nM) loaded with 1, 10, or 100 nM SIINFEKL peptide, respectively, and unloaded spleen cells were labeled with 10 μM chloromethyl-benzoyl-aminotetramethyl-rhodamine (Invitrogen/Molecular Probes). A total of 12 × 106 cells per mouse were injected, consisting of a mixture containing each target cell population. Inguinal LNs draining the immunization site were collected 24 h after injection of target cells. Presence of viable injected target cells was determined using exclusion by 7-aminoactinomycin D. Percentage killing was calculated using the formula as described (24).
Tumor induction
The OVA-expressing thymoma EG7 was obtained from the American Type Culture Collection. The B16-OVA melanoma was a gift from Dr. J. Hess (Erlangen, Germany). PancOVA was generated from the pancreatic cancer line Panc02 (18). For tumor induction, 106 tumor cells in 100 μl PBS were injected s.c. into the flank of age-matched vaccinated and unvaccinated mice. Tumor size was determined using a caliper, and mice were culled when the tumor size reached 200 mm2.
Statistical analysis
A Student t test was applied to reveal significant differences between groups. A p value < 0.05 was accepted as the level of significance.
Results
Subcutaneous injection of an OVA/ISCOM vaccine and ISCOMATRIX adjuvant leads to recruitment and activation of innate and adaptive immune effector cells in VDLNs
To characterize the immunostimulatory effect of ISCOM vaccines on leukocyte populations in vivo, we injected mice s.c. with OVA, the OVA/ISCOM vaccine, or ISCOMATRIX adjuvant without Ag and removed VDLNs for analysis of leukocyte composition and activation marker expression. Both ISCOMATRIX adjuvant and the OVA/ISCOM vaccine induced a visible enlargement of the VDLNs, reaching a maximum after 24 h (Supplemental Fig. 1A). At 24 h the total cell number had increased 5- to 6-fold as compared with VDLNs of mice injected with PBS or OVA alone (Fig. 1A). FACS analysis of VDLN cell suspensions revealed a significant increase of B cell, CD4+ T cell, and CD8+ T cell numbers (Fig. 1B); overall, the CD4/CD8 ratio was not affected (data not shown). Additionally, we observed a dramatic increase in the numbers of NK1.1+CD3− NK cells (13-fold) and NK1.1+CD3+ NKT cells (11-fold) (Fig. 1C). Interestingly, VDLNs contained a population with high sideward light scatter property and coexpression of Ly6G+ and CD11b+, which was absent in control LNs (Fig. 1D). These cells could be identified as granulocytes using H&E staining (Supplemental Fig. 1B). Importantly, this immune stimulatory effect was limited to the VDLNs, since contralateral LNs and spleens did not differ in cell number or composition from control animals (data not shown).
Next, we characterized the activation status of innate and adaptive immune effector cells in VDLNs by measuring CD69 and intracellular IFN-γ expression. The OVA/ISCOM vaccine induced significant CD69 upregulation on B cells, CD4+ T cells, and CD8+ T cells (Fig. 2A). Additionally, we observed upregulation of CD69 expression and IFN-γ production by NK cells and NKT cells (Fig. 2B). Overall, we observed no difference between ISCOMATRIX adjuvant alone and the OVA/ISCOM vaccine with regard to recruitment of leukocyte populations and their activation, indicating that these effects were mediated by the immunostimulatory capacity of ISCOMATRIX adjuvant.
The OVA/ISCOM vaccine recruits DCs to VDLNs and induces DC activation in vivo
DCs comprise a heterogeneous family of APCs. Lymphatic tissues in mice contain three main DC types, which express CD11c and can be discriminated as CD8α+CD205+, CD4+CD8α−, and CD4−CD8α− DCs. In skin-draining LNs, two additional DC populations, CD205+Langerin+ DCs and CD205+CD11b+ interstitial DCs, are found. Of these, CD8α+ DCs are highly specialized in cross-presentation and CTL induction (25). To assess the influence of the OVA/ISCOM vaccine and ISCOMATRIX adjuvant on DCs, we quantified CD11c+ DCs in LN preparations of vaccinated mice. A significant increase of CD11c+ DCs was observed after 24 h for both ISCOMATRIX adjuvant and the OVA/ISCOM vaccine. Because of the low number of DCs in VDLNs, we limited our analysis to CD8α+ and CD8α− DC subsets. The increase of total DCs was mainly due to recruitment (or retention) of CD8α− DCs, which made up ∼90% of CD11c+ DCs in the VDLNs. Additionally, the vaccine induced a 4- to 5-fold increase in CD8α+ DCs (Fig. 1E).
The importance of DC activation for CTL induction has been demonstrated in DC-based therapy studies (3, 4). In this respect, Th1 responses are required, which are dependent on the production of Th1-driving cytokines by DCs, such as IL-12 (26). Thus, we analyzed expression of the costimulatory molecule CD86 and IL-12 production by DCs in VDLNs. Both the OVA/ISCOM vaccine and ISCOMATRIX adjuvant induced a significant upregulation of CD86 by DCs in vivo. This effect was most pronounced for CD8α+ DCs (Fig. 2C) and was similarly effective as the TLR9 ligand CpG-ODN1826 (data not shown). Additionally, a fraction of CD8α+ DCs in VDLNs produced IL-12, whereas CD8α− DCs expressed no IL-12p40 above background levels (Fig. 2C).
The OVA/ISCOM vaccine targets Ag to DCs in VDLNs and induces effective cross-presentation for CTL activation
To assess in vivo whether the OVA/ISCOM vaccine targets Ag to DCs, mice were injected with Alexa 488-labeled OVA, either as soluble protein or formulated as an OVA/ISCOM vaccine. VDLNs and contralateral LNs were removed after 24 h and Ag uptake by CD11c+ DCs was studied by flow cytometry. Analysis revealed Ag uptake by ∼15% of CD8α− DCs and 20% of CD8α+ DCs in the VDLNs. Interestingly, uptake of the ISCOM vaccine was more efficient than uptake of soluble protein by CD8α− DCs. A similar trend was observed for CD8α+ DCs, but this was not statistically significant (Fig. 3A).
To evaluate the capacity of DCs to cross-present the Ag in vivo, we transferred CFSE-labeled T cells from OT-1 mice, which recognize the H2-Kb–restricted epitope SIINFEKL, into mice vaccinated with either soluble OVA (30 μg) or the OVA/ISCOM vaccine (containing 0.3 μg OVA). OT-1 T cell proliferation and IFN-γ production were assessed after 3 and 10 d in the VDLNs by flow cytometry. As shown in Fig. 3B, the OVA/ISCOM vaccine induced a potent T cell proliferative and IFN-γ response. In contrast, even 100-fold higher concentrations of soluble protein induced only minor T cell proliferation. Interestingly, cross-presenting capacity of APCs in VDLNs was lasting for a period of up to 10 d after vaccination. To assess which DC population was most efficient in CD8+ T cell activation, we isolated LN-resident CD8α+DEC205+ DCs, LN-resident CD8α−DEC205− DCs, and migratory CD8α−DEC205+ DCs from VDLNs by FACS and cocultured the DC subsets with CFSE-labeled OT-I T cells. This experiment revealed that CD8α+ DCs were most efficient in cross-presenting the OVA/ISCOM vaccine in vivo (Fig. 3C). Cross-presentation efficacy was further studied with splenic CD8α+ DCs pulsed with either soluble OVA or OVA/ISCOM in vitro. These experiments confirmed highly effective cross-presentation of the vaccine Ag by this DC subset and further demonstrated a >100-fold Ag-sparing effect as compared with soluble Ag (Fig. 3D).
The OVA/ISCOM vaccine induces high levels of IL-1β and IL-6 as well as a mixed Th1/Th2 cytokine profile in VDLNs
To evaluate the cytokine response mediated by the ISCOM vaccine, we vaccinated mice and removed the VDLNs after 6, 12, and 24 h. VDLNs of mice injected with either PBS or OVA served as controls. Concentrations of IL-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p40, IL-12p70, GM-CSF, and TNF-α were measured by multiplex assay. Strikingly, high levels of IL-1β and IL-6 were produced in VDLNs as early as 6 h after injection of the vaccine, indicative of an early inflammatory response (Fig. 4A). Additionally, the vaccine induced the production of GM-CSF and IL-12p40, as well as low levels of IL-4 and IL-5, indicative of a mixed Th1/Th2 type response. No significant changes in IL-2, IL-10, IL-12p70, or TNF-α production were observed. To confirm that IL-1β was secreted in its bioactive form (17 kDa) we performed Western blot analysis of VDLN lysates. As shown in Fig. 4B, p17 was detected in VDLNs but not control LNs (PBS). IL-1β secretion in response to the ISCOM vaccine was confirmed in LPS-primed DCs in vitro and could be blocked with z-YVAD, an inhibitor of caspase-1 (Fig. 4C). Thus, ISCOM vaccines induce caspase-1–dependent cleavage of pro–IL-1β to active IL-1β in DCs.
The OVA/ISCOM vaccine induces priming of CTL capable of tumor cell killing
We analyzed the vaccine-induced immune response using a prime-boost regimen in which mice were vaccinated twice with OVA/ISCOM vaccine (0.3 μg OVA) or soluble OVA (30 μg) in a weekly interval. To compare the potency of the ISCOM vaccine formulation with other adjuvant systems, we also vaccinated mice with OVA (30 μg) together with either TLR ligands (Pam3-Cys for TLR3, LPS for TLR4, CpG-ODN1826 for TLR9), IFA, or alum. On day 14, priming of OVA-specific CTLs was assessed by measuring the frequency of splenic CD8+ T cells specific for the H2-Kb–restricted epitope SIINFEKL by IFN-γ ICS. The OVA/ISCOM vaccine induced highly efficient cross-priming of OVA-specific CTLs with an average CTL frequency of 8.5% of total CD8+ T cells (Fig. 5A). Neither OVA alone or in combination with Pam3-Cys, IFA, or alum induced a detectable CTL response. Of the TLR ligands, CpG-ODN1826 and LPS induced the best CTL responses with a frequency of 1.8 and 0.7% of CD8+ T cells, respectively. To analyze the role of DCs in CD8+ T cell cross-priming, we vaccinated irradiated mice reconstituted with bone marrow from CD11c-DTR mice, in which CD11c+ DCs can be ablated by injection of DT (22). The vaccination protocol and depletion efficacy are shown in Supplemental Fig. 2B. Importantly, non-CD11c–expressing APCs such as radio-resistant Langerhans cells, B cells, and monocytes are not depleted and can still perform APC functions after DT treatment. As shown in Fig. 5B, depletion of CD11c+ DCs almost completely abolished CTL induction. To study the role of Langerhans cells and Langerin+ dermal DCs in T cell priming, we made use of Langerin-DTR mice, in which Langerin+ cells can be depleted by repeated injections of DT (the vaccination protocol and depletion efficacy are depicted in Supplemental Fig. 2A). Depletion of Langerin+ APCs had no impact on CTL priming or cytotoxic activity in vivo, indicating that these DC subsets are dispensable for ISCOM vaccine-mediated CTL induction (Fig. 5C, 5D). These data strengthen the hypothesis that CTL priming by ISCOM vaccines is predominantly mediated by conventional DCs, with other APCs playing a less important role.
The functional activity of the CTL was studied in an in vivo cytotoxicity assay using adoptively transferred SIINFEKL-pulsed and CFSE-labeled splenocytes. Specific target cell lysis was 80 and 60% on days 10 and 35 after the second vaccination, respectively, indicative of T cell memory induction (Fig. 6A). No significant killing was observed in mice vaccinated with soluble OVA. Moreover, vaccinated mice were protected against challenge with OVA expressing tumor cells, such as EG-7 thymoma, PancOVA pancreatic cancer, or B16-OVA melanoma cells (Fig. 6B). Vaccinated animals remained tumor free during an observation period of 100 d after tumor induction. Mice did not develop tumors after a second tumor challenge, indicative of long-term memory induction (data not shown).
Discussion
ISCOMATRIX adjuvant-based vaccines hold promise for generating effective CTL responses against tumors (19). In this study, we investigated mechanisms of action of ISCOMATRIX adjuvant and an OVA/ISCOM vaccine. We used the ISCOM vaccine rather than the ISCOMATRIX vaccine to be sure that the Ag and adjuvant were associated, thereby enabling labeling of the Ag to track the formulation. We found that ISCOM vaccines 1) induce potent immune activation in VDLNs, 2) effectively target Ag to DCs in vivo, 3) induce DC maturation and IL-12 production in vivo, 4) facilitate Ag cross-presentation by CD8α+ DCs, and 5) mediate a cellular immune response leading to tumor protection and T cell memory induction.
Immune activation by an ISCOMATRIX vaccine has been demonstrated by lymphatic cannulation in sheep with the transient appearance of proinflammatory cytokines in efferent lymph, including IFN-γ, IL-6, IL-8, and IL-1β (27). Consistent with this report, we found high levels of bioactive IL-1β and IL-6 in VDLNs in mice. Additionally, we found increased levels of IL-5 and IL-12, indicative of a mixed Th1/Th2 response. IL-1β is a proinflammatory cytokine that mediates recruitment of leukocytes to sites of inflammation and the secretion of other proinflammatory cytokines, such as IL-6. The secretion of bioactive IL-1β is regulated by caspase-1–activating inflammasome complexes, such as the NLRP3 inflammasome (28). Interestingly, we observed secretion of bioactive IL-1β by DCs incubated with ISCOMATRIX adjuvant in vitro, which occurred in a caspase-1–dependent manner. Assessing the role of different inflammasome types, such as the NLRP3 inflammasome, on the adjuvant properties of ISCOMATRIX adjuvant is the focus of ongoing studies.
The immunostimulatory effect of ISCOMATRIX adjuvant was further demonstrated by the recruitment of activated innate and adaptive effector cells to VDLNs. The increase of total cell composition within VDLNs was predominantly due to recruitment of B cells and T cells, expressing high levels of the early activation marker CD69. Additionally, a marked increase of NK cells and NKT cells producing IFN-γ as well as an influx of granulocytes was observed. A cross-talk between NK cells and DCs has been suggested to play an important role in anti-tumor immunity. Mature DCs can induce IFN-γ expression in NK cells (29). Reciprocally, NK cells have the capacity to induce DC maturation and IL-12 production (30). A similar interplay between granulocytes and DCs leading to DC maturation and enhanced Ag presentation has been reported (31, 32). Indeed, VDLNs contained increased numbers of CD8α+ DCs expressing high levels of costimulatory molecules, MHC class II (data not shown), and IL-12p40. Taken together, these results indicate that ISCOMATRIX adjuvant is a potent immune modifier, linking innate and adaptive immunity and creating a favorable milieu for Th1 type immune responses.
Because CD8α+ DCs are highly specialized in cross-priming CTL responses, tumor vaccines should aim at targeting this DC subpopulation (and the corresponding DC population in humans) in vivo. Whether other DC populations, such as Langerhans cells, can also cross-prime CTLs is controversial (33, 34). In this study, we show that the ISCOM vaccine is effectively taken up by both CD8α+ and CD8α− DCs (including skin-derived migratory DCs) in VDLNs, indicative of effective Ag targeting to DCs in vivo. Cross-presentation was highly effective as demonstrated by the potent proliferative response of adoptively transferred OT-1 T cells. FACS sorting of DC populations in VDLNs revealed that cross-presenting capacity was mainly attributed to CD8α+ DCs, at least during the first 24 h. Moreover, deletion of both Langerhans cells and dermal Langerin+ DCs in Langerin-DTR mice with the DT protocol used in our study (35) did not influence CTL cross-priming. In contrast, deletion of CD11c+ DCs in mice with bone marrow from CD11c-DTR mice almost completely abrogated CTL induction. Taken together, these data indicate that conventional CD8α+ DCs are the main cell type mediating ISCOM vaccine-induced CTL responses. Whether other DC types or even granulocytes, which effectively engulfed the ISCOM vaccine (see Supplemental Fig. 2C), play a role in transporting the vaccine Ag from the injection site to the VDLNs should be investigated in further studies (36, 37). Of note, the efficacy of the ISCOM vaccine in priming CTL responses was striking with an average of 8.5% of Ag-specific CD8+ T cells in peripheral blood after only two vaccinations. Side-by-side comparisons with other adjuvant systems proved the high efficacy of ISCOM vaccines. Even at a 100-fold higher Ag concentration, combinations of OVA with TLR ligands (TLR2, TLR4, or TLR9), IFA, or alum were far less effective in cross-priming a CTL response. Additionally, we could demonstrate effective CTL responses in three different tumor models, in which immunization with an ISCOM vaccine using a prime-boost regimen mediated complete tumor protection against OVA-expressing tumor cells.
In conclusion, ISCOMATRIX adjuvant combines characteristics of a potent immune modifier with effective Ag delivery to DCs in vivo and leads to the induction of a strong cellular immune response. ISCOMATRIX adjuvant-based vaccines have shown promising results in patients with cancer (20). The data presented in this study support our understanding of the underlying cellular mechanism of ISCOM and ISCOMATRIX vaccines and will allow us to enhance vaccine efficacy for inducing effective anti-tumor immune responses.
Acknowledgements
We are grateful to Dr. Gabrielle Belz and Dr. Adele Mount for providing CD11c-DTR mice for vaccination experiments.
Footnotes
This work was supported by grants from the Deutsche Krebshilfe (to M.S.), Deutsche Forschungsgemeinschaft (SCHN 664/3-1 to M.S. and GK1202 to M.S., C.B., D.A., and S.E.), BayImmunet (to C.B. and S.E.), and Austrian Science Fund (P-214780 to P.S.). B.E.C. is a VIDI Fellow of the Netherlands Organization for Scientific Research (917-76-365). This work is part of the doctoral thesis of U.K. at the University of Munich.
The online version of this article contains supplemental material.
References
Disclosures
E.M., A.B., and S.K. are employees of CSL Ltd. The remaining authors have no financial conflicts of interest.