|Home | About | Journals | Submit | Contact Us | Français|
Evaluation of T cell responses to tumor- and pathogen-derived peptides in preclinical models is necessary to define the characteristics of efficacious peptide vaccines. We show here that vaccination with insect cells infected with baculoviruses expressing MHC class I linked to tumor peptide mimotopes results in expansion of functional peptide-specific CD8+ T cells that protect mice from tumor challenge. Specific peptide mimotopes selected from peptide-MHC libraries encoded by baculoviruses can be tested using this vaccine approach. Unlike other vaccine strategies, this vaccine has the following advantages: peptides that are difficult to solublize can be easily characterized, bona fide peptides without synthesis artifacts are presented, and additional adjuvants are not required to generate peptide-specific responses. Priming of antitumor responses occurs within three days of vaccination and is optimal one week after a second injection. After vaccination the antigen-specific T cell response is similar in animals primed with either soluble or membrane-bound antigen, and CD11c+ dendritic cells increase expression of maturation markers and stimulate proliferation of specific T cells ex vivo. Thus, the mechanism of antigen presentation induced by this vaccine is consistent with cross-priming by dendritic cells. This straightforward approach will facilitate future analyses of T cells elicited by peptide mimotopes.
Identification of MHC class I and MHC class II binding epitopes have expedited the use of peptides in immunotherapy. Peptide vaccine strategies target T cells with fine specificity and, in combination with the appropriate adjuvants, generate immune responses to pathogens and cancers. Peptides in combination therapies may be key in the next generation of vaccines.
Many studies have employed insect cells infected with baculoviruses (BV) for production of proteins used in vaccines (1–3). Due to the large viral genome and strong promoters, BV vectors accommodate large gene inserts (>1 Kb) and produce high yields of mammalian proteins (4). Furthermore, post-translational modifications, such as glycosylation and phosphorylation, in insect cells are similar to mammalian processes, allowing expression of proteins that biochemically resemble those of mammalian origins (5). For example, Spodoptera frugiperda (Sf9) and High Five insect cells infected with recombinant BV produce soluble immunogenic viral proteins and viral-like particles from HIV and Foot and Mouth Disease Virus (FMDV) for use in vaccines (6, 7). Both serological (8) and cellular responses (2) are elicited by purified HIV proteins produced by BV-infected insect cells, which protect animals against subsequent viral challenge. Although vaccination with protein produced by BV-infected insect cells induces antigen-specific immune responses, this vaccination strategy requires protein purification and appropriate adjuvants.
Since the BV polyhedron promoter is not active (9–11) and BV cannot replicate in mammalian cells (12), injection of recombinant BV may provide effective and safe means for delivery of vaccines. Recombinant BV expressing immunogenic proteins linked to the transmembrane domain of gp64 for viral surface expression elicit antigen-specific responses (13–15). However, recombinant BV are inactivated by complement proteins in vivo (16) and may be damaged during purification processes, particularly by ultracentrifugation (17).
Injection of insect cells infected with recombinant BV is an attractive method for vaccine delivery because it combines benefits from both the protein and viral vaccines. It has been shown that these vaccines elicit humoral immune responses to surface-expressed viral antigens (18). For example, vaccination with infected insect cells expressing FMDV antigen elicits seroneutralizing antibodies resulting in protection from viral challenge (7). The recombinant proteins are produced in culture where complement proteins do not interfere with antigen production. In addition, the antigen can be quantified prior to injection and preparation of infected insect cells for vaccination requires only low-speed centrifugation. Thus, we hypothesized that vaccination with insect cells infected with BV encoding tumor-specific antigens would be a promising technique for priming specific CD8+ T cell responses.
Peptide-MHC complexes and peptide-MHC libraries used for the discovery of novel peptide antigens are successfully produced by insect cells infected with recombinant BV (19–23). These peptide libraries are screened for binding to soluble TCR and activation of T cells in vitro prior to testing the peptides in vivo. Peptides produced in the BV peptide-MHC library are soluble and not easily oxidized, which permits screening of all amino acid residues, including cysteine and tryptophan. In theory, peptide epitopes or peptide mimotopes identified using this library system may regulate the T cell response and thus the disease progression in autoimmunity, cancer, and infectious diseases [reviewed in (24)].
We are using BV peptide-MHC libraries to identify novel cancer mimotopes, or mimics of tumor peptides, that stabilize the peptide-MHC/TCR complex and elicit T cells that cross react with the tumor antigen (23). Like the peptide mimotopes we have identified, most mimotopes used in clinical trials of cancer vaccines have alterations in the MHC-anchor residues [reviewed in (25)]. We are characterizing mimotopes that improve antitumor immunity to the CT26 mouse colon carcinoma, specifically to the immunodominant tumor antigen AH1 (gp70423–431) (26), restricted by the MHC class I molecule H-2Ld. We previously showed that antitumor activity is improved by vaccinating with mimotope-liposome complexes (27) or mimotope-loaded DCs (28). Because some mimotopes are insoluble in water, sensitive to oxidation, and cannot be characterized using these methods, we developed a vaccine using infected insect cells expressing peptide-MHC molecules.
We demonstrate here that vaccination with insect cells infected with recombinant BV encoding peptide-MHC complexes generates peptide-specific cytotoxic T cell responses and when the appropriate peptide is used, protects mice from subsequent tumor challenge. Our results indicate that the infected insect cells activate antigen-presenting cells in vivo, which effectively present the expressed peptides to T cells. This vaccination strategy is advantageous for the following reasons: it ensures the bone fide peptide is presented, it does not require adjuvants in addition to those produced by BV and insect cells, it greatly reduces the cost of in vivo studies by eliminating the need to synthetically generate peptides, and it expedites the direct evaluation of peptides identified in BV peptide-MHC libraries.
Sf9 and High Five insect cells [(22), Invitrogen, Carlsbad, CA], and CT26 tumor cells (29) were cultured as described. Splenocytes from vaccinated mice were expanded in vitro with AH1 peptide and IL-2 as described (27). Dendritic cells were prepared from collagenase-digested spleens or mesenteric lymph nodes for flow cytometric analyses or in vitro proliferation assays as described [200 µg/mL Collagenase D (Roche, Nutley, NJ) and 40 µg/mL DNAse-I (Sigma, St. Louis, MO) (30)]. Mesenteric lymph node cells were harvested 0 (unvaccinated), 2, 8, 16, 24, or 48 h after vaccination with infected insect cells for flow cytometric analyses. CD11c+ cells were isolated for in vitro proliferation assays 24 h after vaccination using a biotinylated CD11c-specific monoclonal antibody and anti-biotin microbeads. Labeled cells were separated using LS MidiMacs columns according to the manufacturer’s protocol (Miltenyi, Bergisch Gladbach, Germany).
CT-TCR transgenic (CT-TCR Tg) mice expressing the TCR from the Vβ8.3/Vα4.11 T cell clone (28) were generated by inserting the TCR α and β genes into shuttle vectors (31), which were subsequently injected into embryos of (SJLXB6)F1 at the University of Pennsylvania Transgenic Facility, and backcrossed to BALB/c 12 generations. Because of low TCR expression, these mice were bred onto a Rag2-deficient background [(C.12956(B6)-Rag2tm1fwaN12), Taconic, Hudson, NY]. Six- to eight-week-old female BALB/c were purchased from the National Cancer Institute/Charles River Laboratories. All animal protocols were reviewed and approved by the Institutional Animal Care and Use Committee of National Jewish Medical and Research Center.
Sequence encoding H-2Ld with either a peptide tag for biotinylation by the enzyme BirA [LdBirA (27)] or the transmembrane domain from gp64 [LdTM (22, 32)] was inserted into a modified pBacp10pH baculovirus expression vector downstream of the p10 promoter (20). Sequence encoding mouse beta-2-microglobulin with covalently-linked peptides [AH1: SPSYVYHQF (26); 39: MNKYAYHML (27); 15: MPKYAYHML (27); βgal: TPHGAGRIL (33); WMF: SPTYAYWMF (23)] was inserted downstream of the pH promoter. The constructions were introduced into BV using the standard homologous recombination method (22). AH1 peptide-loaded LdBirA used for the ELISA standard was purified from supernatants of infected High Five insect cells over an affinity column using an antibody specific for H-2Ld (28.14.8s, ATCC, Manassas, VA). Protein-containing fractions were concentrated and separated on a Superdex-200 sizing column. AH1 peptide (Macromolecular Resources, Ft. Collins, CO) was added to the 57 kD fraction in 5-fold molar excess. Fluorescent tetramer was prepared as described (27).
Soluble TCR was constructed by inserting the TCR-encoding V-region gene fragments from CT-Ig (28) into a modified pBacp10pH BV expression vector (23). CT-TCR soluble protein was purified from supernatants of infected High Five insect cells over an affinity column using an antibody specific for TCR Cβ (Ham-597, ATCC, Manassas, VA) and a Superdex-200 sizing column. Purified CT-TCR was multimerized with a biotinylated anti-TCR Cα specific antibody (ADO-304) and streptavidin-AF647 (Invitrogen Molecular Probes, Carlsbad, CA) as previously described (22). Antibodies specific for H-2Ld (28.14.8s), CD80 (16-10A1, eBioscience, San Diego, CA), CD86 (GL1, BD Pharmigen, San Jose, CA), MHC class II (M5/114.15.2, BD Pharmigen, San Jose, CA), CD11c (N418, BD Pharmigen, San Jose, CA), CD11b (M1/70, eBioscience, San Diego, CA), CD8β (2.43, ATCC, Manassas, VA), Vβ8.3 (CT-8C1, BD Pharmigen, San Jose, CA), IFN-γ (XMG1.2, eBioscience, San Diego, CA), and the compounds 7-Amino-Actinomycin D (7-AAD, Sigma, St. Louis, MO) and carboxyfluorescein diacetate succinimidyl ester (CFSE, Invitrogen Molecular Probes, Carlsbad, CA) were used for flow cytometric analyses. Mice were depleted with intraperitoneal (I.P.) injections of antibody specific for CD8β (53.6.72, ATCC, Manassas, VA) 3 days (500 µg) and one day (250 µg) prior to vaccination. Depletion was maintained with weekly injections of 250 µg of antibody and was confirmed by flow cytometry (≥ 99.8% depletion prior to vaccination and ≥ 79% depletion prior to tumor challenge, data not shown).
3×107 Sf9 insect cells were cultured in T175 flasks in complete Grace’s Insect medium (Invitrogen, Manassas, VA) containing 10% fetal calf serum (Atlanta Biologicals, Lawrenceville, GA), 1% F-68 detergent (Invitrogen, Manassas, VA), and 1% Antibiotic-Antimycotic (Invitrogen, Manassas, VA). The BV titer was determined using a limiting dilution assay. When a multiplicity of infection of 2 units/cell is used for each infection, consistent infection efficiencies and insect cell death rates (20% by day 3) are obtained. Infected Sf9 insect cells were incubated for three days, harvested by centrifugation at 1000 × g for 5 minutes, and washed 3 times with HBSS (Mediatech, Inc, Hendon, VA).
Infected Sf9 insect cells were resuspended in HBSS and 5×106 cells were injected I.P. on days 0 and 7. The number of responding antigen-specific T cells is similar following intravenous and subcutaneous injection. Splenocytes or PBMCs were harvested for flow cytometric analyses on days 10, 14, and 17. Statistical analyses were performed with Prism version 4.0, GraphPad Software, using an unpaired 2-tailed Student’s t test. A p value of less than 0.05 was considered statistically significant.
Whole cell lysates were prepared by incubating infected Sf9 insect cells in lysis buffer and protease inhibitors as previously described (34) at a concentration of 1×107 cells/mL for 4 h at 4°C. H-2Ld was immunoprecipitated from whole cell lysates with the 28.14.8s antibody and Protein A sepharose beads (35). Precipitated proteins were separated by SDS-PAGE (5–20% Tris-HCl, Biorad, Hercules, CA) under reducing conditions using a standard protocol and stained with Coomassie Blue.
Mice were vaccinated with 39-LdTM-infected or uninfected insect cells on days 0 and 7. 30 d after the last vaccination, target cells (BALB/c splenocytes) were incubated with either βgal or AH1 peptide (10 µg/mL) for 2 h at room temperature. Cells were washed and labeled with 0.2 µM or 2 µM CFSE, respectively, and injected intravenously. Splenocytes were harvested 20 h later and the number of CFSE+ cells in each peak was determined [% specific killing = 1 – % survival; % survival = (# of AH1 targets remaining) / (# of βgal targets remaining)]. Groups were compared using Prism version 4.0, GraphPad Software, by an unpaired 2-tailed Student’s t test. A p value of less than 0.05 was considered statistically significant.
For in vivo proliferation assays, mice were vaccinated as described above with either 39-LdTM-infected or uninfected insect cells on day -7, -3, or -1. 1×107 splenocytes from CT-TCR Tg mice were labeled with 10 µM CFSE and transferred into vaccinated mice on day 0. Three days later CFSE dilution of transferred Vβ8.3+ CD8+ splenocytes was determined by flow cytometry. Background proliferation was determined in mice vaccinated with uninfected insect cells.
For in vitro proliferation assays, 5×105 CFSE-labeled splenocytes (labeled as above) from CT-TCR Tg mice were incubated in 96-well plates at 37°C with increasing concentrations of soluble peptide or 1×105 CD11c+ splenocytes pre-incubated with 100 µg/mL peptide or from vaccinated mice in complete medium (27). Cells were harvested 3 days later and CFSE dilution of 7-AAD− CD8+ cells was analyzed by flow cytometry. In vitro proliferation assays using a T cell clone expressing the CT-TCR were performed as previously described (27). The T cell clone was incubated at a 5:1 ratio with insect cells that express ICAM and B7 costimulatory molecules (22) and infected with BV encoding peptide-MHC.
For tumor protection experiments, mice were injected I.P. with 5×106 peptide-LdTM infected Sf9 insect cells on days –14 and –7. On day 0, mice were injected subcutaneously in the left hind flank with 5×104 CT26 tumor cells (26). Tumor-free survival was assessed by palpation of the injection site. Once tumors were palpable, they always proceeded to 100 mm2 without shrinking. When a tumor reached 100 mm2, the mouse was no longer considered tumor-free as indicated on the Kaplin-Meier plot and it was sacrificed. All mice are represented in the Kaplin-Meier plot. Tumor-free survival was analyzed by Kaplan-Meier survival plots and statistical significance was analyzed with Prism version 4.0, GraphPad Software, using the log rank test.
For tumor treatment experiments, mice were injected with 5×104 CT26 tumor cells, and vaccinated with 5×106 infected insect cells 2, 5, 8, 11, and 14 days later. Tumors were measured every two days and groups were compared statistically on individual days using Prism version 4.0, GraphPad Software with an unpaired 2-tailed Student’s t test. A p value of less than 0.05 was considered statistically significant. Differences in tumor size of mice injected with 39-LdTM relative to unvaccinated or βgal-LdTM were statistically significant after day 9. The average tumor size of the indicated number of mice is plotted.
To ensure that insect cells infected with BV produce antigen recognized by cognate TCR, we generated BV encoding peptide-Ld molecules. The transmembrane domain of BV gp64 was inserted downstream of H-2Ld for surface expression [LdTM, (6, 21–23)] and peptides were tethered to the β2M molecule via a glycine-rich linker. We inserted either the CT26 tumor antigen (the AH1 peptide) or the negative control βgal peptide, which binds to H-2Ld but is not recognized by the AH1-specific TCR (CT-TCR). To ensure that the cognate TCR recognizes peptides produced in insect cells with similar relative affinity as synthetic peptides, we also generated BV encoding H-2Ld covalently linked to previously studied peptide mimotopes with changes in the MHC-binding residues of different affinities (27). The CT-TCR binds to mimotope 39-Ld with an intermediate affinity and to mimotope 15-Ld with a high affinity, referring to the peptide-MHC/TCR interaction, as determined by surface plasmon resonance (27).
Three days after infection with BV, insect cells were stained with antibodies specific for H-2Ld bound to β2M and peptide (28.14.8s) and soluble TCR multimer [predicted octamer, CT-TCR (23)]. As shown previously, the amount of MHC expression on the surface of the insect cells correlates with the extent of viral infection (22) and is consistent between experiments. Thus, TCR staining within a given intensity of MHC staining, represented by the thin gate in Figure 1a, can be compared between samples because the insect cells are infected similarly.
As expected, similar amounts of H-2Ld protein was detected on the surface of insect cells infected with BV encoding all of the peptide-LdTM constructions. βgal-LdTM was detected with the H-2Ld antibody, but not with the CT-TCR, demonstrating specificity of the CT-TCR reagent (Figure 1a). Importantly, as shown with other BV-encoded peptide-MHC complexes (22), the CT-TCR fluorescence intensity directly correlated with its affinity for the peptide-MHC molecules (Figure 1b). Specifically, 15-LdTM stained most intensely with CT-TCR followed by 39-LdTM, and AH1-LdTM.
These results show that peptide-Ld complexes produced by insect cells are processed and folded to resemble those produced in mammalian cells. Furthermore, the binding properties of the covalently linked peptides directly correlate with the binding properties of soluble peptides, suggesting that insect cell-produced peptide-Ld complexes bind antigen-specific T cells. Finally, these results confirm the results of Crawford et al (22): binding affinity of the peptide-MHC/TCR interaction can be readily analyzed using these BV constructions.
We hypothesized that the infected insect cells provide both an antigen-specific signal and adjuvant from the combination of BV and foreign insect cells. To determine if infected insect cells induce antitumor responses in vivo, we analyzed AH1-specific CD8+ T cells following injection of these cells. To produce this “vaccine,” we infected Sf9 insect cells for 3 days, harvested and washed the cells, then injected them intraperitoneally. Although the insect cells were washed, remaining free virus and dead insect cells were also included.
We previously showed that vaccination with the intermediate affinity mimotope 39 in liposomes elicits tumor-specific T cells and protects about 50% of mice from tumor growth (27). Although the high affinity mimotope 15 elicits tumor-specific T cells, it does not protect mice from tumor growth. Thus, for simplicity we used insect cells infected with 39-LdTM throughout the following experiments. As shown in Figure 2a, AH1-specific (tumor-specific), but not βgal-specific, T cells were elicited after injection of insect cells infected with BV encoding mimotope 39 as determined by tetramer staining. AH1-specific T cells were not elicited with the βgal peptide vaccine, the negative control.
We next determined the optimal number of insect cells needed to prime the T cell response and calculated the corresponding amount of peptide-Ld produced in infected insect cells by ELISA. Mice were injected with increasing numbers of 39-LdTM infected Sf9 insect cells on days 0 and 7. Splenocytes were harvested on day 14 and cultured for one week. Insect cells infected with 39-LdTM, elicited CD8+ AH1-Ld tet+ (tumor specific) T cells in a dose-dependent manner (Figure 2a). This increase was most evident 7 days after the second injection (day 14), as determined by the average frequency of tetramer-positive PBMCs on days 10, 14, and 17 (Figure 2b). Although there are stochastic differences in the frequency of AH1-Ldtet+ T cells, the increase in the percentage of these T cells at day 14 is significant different from day 17. Significant expansion of AH1-Ld tet+ T cells was not detectable prior to the second injection (data not shown). There was minimal background βgal-tetramer staining in these experiments (≤0.2%, Figure 2a) and no adverse side effects were observed in the injected mice.
To determine the amount of antigen delivered by this vaccine, we compared the amount of H-2Ld protein in the infected insect cell vaccine to a standard curve of purified H-2Ld by ELISA using conformation-specific antibodies. The vaccine was prepared for the ELISA as it was for injection. We calculated that insect cells infected with 39-LdTM BV produce approximately 2 pg, or 2×107 peptide-MHC molecules per cell (data not shown). Vaccination with 5×106 infected insect cells therefore delivers 10 µg (standard deviation = 1.82 µg, n = 3) of peptide-MHC complexes or 200 ng of peptide. This vaccination strategy delivers more MHC molecules than an exosome-based vaccine in which mice are vaccinated with up to 1×1010 molecules of MHC per mouse (36) and less than 10 µg of peptide used in the peptide-liposome vaccine (27).
We next determined if vaccine-elicited T cells were functional in an in vivo killing assay (Figures 2c and d). Splenocytes from BALB/c mice were incubated with AH1 or βgal peptides and labeled with a high or low concentration of CFSE, respectively. These labeled splenocytes were transferred into vaccinated BALB/c mice 30 days following the second injection of 39-LdTM-infected insect cells. AH1 peptide-loaded target cells were specifically eliminated in mice vaccinated with 39-LdTM-infected insect cells, but not with uninfected insect cells (Figure 2d). The number of βgal-loaded targets remained similar in both samples. These results indicate that 39-LdTM-infected insect cells elicit effector T cells that specifically kill antigen-loaded target cells in vivo. Thus, the same viral constructions can be used for in vivo and in vitro analyses.
To analyze T cell responses to this vaccine and ultimately to design mimotopes to tumor-associated antigens, we derived transgenic mice that express the TCR from the CT T cell clone (28), a clone that recognizes the AH1 peptide restricted by H-2Ld. Like many other T cells that recognize tumors, this T cell clone recognizes a self antigen, and therefore is subject to negative selection in the thymus and peripheral tolerance after leaving the thymus. We developed this new transgenic mouse model, rather than using an established transgenic strain, such as the OT-1 mice, to better mimic T cell tolerance encountered by tumor vaccines. We backcrossed the transgenes onto BALB/c mice for 12 generations and then crossed the transgenes onto Rag2-deficient mice. Approximately 90% of the T cells in the thymus of these transgenic mice are co-receptor-negative (data not shown) indicating 1) strong negative selection of the T cells during development, 2) a developmental block in the T cells at the double negative stage, as the CT-TCR is specific for a self peptide derived from an endogenous retroviral gene product, gp70, and/or 3) the Ig enhancer used to drive gene expression is suboptimal (31).
As in other models of self tolerance, some T cells escape negative selection and are found in the periphery (37). Some of the peripheral Vβ8.3+ T cells from the transgenic mice express CD8 molecules and these T cells are functional as determined by tetramer binding and other assays (Figure 3). The co-receptor-negative cells bind a Vβ8.3 antibody (Figure 3b), but they do not all bind to AH1-Ld tet (Figure 3a), suggesting that T cells lacking co-receptor may require a higher affinity peptide to form a complex. Consistent with this possibility, more co-receptor negative T cells bind 39-Ld tet than AH1-Ld tet (data not shown). Eighty to 90% of the CD8+ T cells proliferated when incubated with 10 nM peptide 39 (Figure 3d). Few of the T cells express CD4 molecules and the remaining T cells are co-receptor-negative (CD4−/CD8−, Figure 3b). The CT-TCR Tg Rag mouse produced functional antigen-specific T cells as determined by production of IFN-γ and proliferation to a range of peptide concentrations (Figures 3c and d). Thus, we determined that these T cells may be used to monitor antigen-specific T cell responses in adoptive transfer assays and other assays to assess tumor-specific T cell responses.
Antigen persistence, or half-life, directly correlates to the potency of a vaccine (38). To determine the persistence of antigen following injection of the BV-infected insect cell vaccine, we analyzed proliferation of transferred splenocytes from CT-TCR Tg Rag mice. BALB/c mice were vaccinated with 39-LdTM-infected or uninfected insect cells, and 1×107 CFSE-labeled transgenic splenocytes were transferred intravenously 1, 3, or 7 days after vaccination. Splenocytes were harvested 72 h after transfer and proliferation of CD8+/Vβ8.3+/CFSE+ cells was determined by flow cytometric analysis (Figure 3e). Vaccination with 39-LdTM-infected insect cells one day prior to transfer induced proliferation of a significant number of transgenic T cells (37%). Proliferation in mice vaccinated 3 days prior to transfer was reduced (16%), although it was significantly higher than in mice vaccinated with uninfected insect cells. Vaccination 7 days prior to transfer did not induce proliferation of transgenic T cells, indicating that most of the antigen is cleared between 3 and 7 days after vaccination. The persistence of infected insect cells is similar to vaccination with antigen and IFA, which are cleared 6–8 d following injection (39).
Although recombinant BV that express both peptides and MHC molecules is convenient and effective for both in vitro and in vivo analyses, we wanted to determine whether antigen is presented directly by insect cells, cross-presented by APCs, or presented using a novel mechanism. To determine if insect cells directly present peptide-Ld to T cells in vivo, we compared vaccination with insect cells expressing membrane-bound peptide-Ld (39-LdTM) to non-membrane-bound peptide 39 (39-LdBirA and peptide-β2M). The BV encoding 39-LdBirA is identical to 39-LdTM but encodes a BirA peptide tag rather than the transmembrane domain of gp64. The sequence encoding H-2Ld was removed from these BV to produce 39-β2M. Insect cells infected with membrane- and non-membrane-bound peptide 39 produce a similar amount of protein, as detected by immunoprecipitation (Figure 4a) and ELISA (data not shown) of whole insect cell lysates 3 days after infection. Although 39-β2M molecules may be detectable in whole cell lysates (Figure 4a), they are not detectable by immunoprecipitation or ELISA as these assays are specific for H-2Ld. As expected, both 39-LdBirA and 39-β2M are not detectable on the surface of infected insect cells with the H-2Ld antibody (Figure 4b), confirming that these infected insect cells do not present MHC-restricted antigens.
If insect cells present antigens directly to T cells, we would expect antigen-specific responses to the insect cells expressing the membrane-bound antigen and not to the insect cells expressing non-membrane-bound antigen. If antigens are cross-presented by APCs, then we would expect little difference between the specific T cell response to insect cells expressing membrane- and non-membrane-bound antigen. We vaccinated mice with insect cells infected with BV expressing 39-LdTM, 39-LdBirA, or 39-β2M and analyzed the frequency of CD8+ AH1-Ld tet+ T cells in the spleen. Insect cells infected with membrane- or non-membrane bound peptide induced a similar number of AH1-specific T cells (Figure 4c). In addition, surface expression on insect cells infected with 39-LdTM that also express costimulatory molecules ICAM-1 and B7.1, facilitators of direct antigen presentation and T cell priming, did not increase the number of responding AH1-specific T cells (data not shown). These results suggest that infected insect cells do not present antigen directly to T cells, but are processed by APCs, which present the peptides in the context of H-2Ld to CD8+ T cells.
To determine if the insect cell vaccine activates APCs as expected, we examined surface markers on DCs after vaccination. Mice were vaccinated with 39-LdTM-infected insect cells and DCs from the draining mesenteric lymph nodes were characterized over time. DCs were stained with antibodies against the DC subset markers CD11c, CD8, and CD11b, and the maturation markers MHC class II, CD80, CD86, and CD70. We observed an increase in the expression of MHC class II and the costimulatory molecules CD80 and CD86 in both DC populations (Figure 5a). Expression of CD70, a molecule expressed by activated DCs that was recently shown to be necessary for optimal T cell stimulation (40), also increased (Figure 5a). Similar results were obtained for DC populations in the spleen (data not shown). Although the vaccine stimulated both CD8+ CD11c+ and CD11b+CD11c+ cells, upregulation of the costimulatory molecules CD80 and CD86 was more pronounced in the CD8+ CD11c+ subset, suggesting that these cells respond more vigorously to the vaccine. The infected insect cell vaccine induces maturation of DCs, particularly the CD11c+ CD8+ DCs, consistent with a function in cross-presentation (41).
To confirm that antigens from infected insect cell vaccines can be cross-presented by DCs, we determined if DCs isolated from mice vaccinated with 39-LdTM-infected insect cells induce proliferation of CT-TCR Tg T cells ex vivo. Mice were vaccinated with 39-LdTM- or βgal-LdTM-infected insect cells and spleens were harvested 24 h later. Purified CD11c+ splenocytes from vaccinated mice (average 88% CD11c+ MHC class II+) were incubated with CFSE-labeled transgenic T cells for three days ex vivo. Like transgenic T cells incubated with DCs exogenously loaded with peptide 39, transgenic T cells proliferated when incubated with CD11c+ cells from mice vaccinated with 39-LdTM-infected insect cells (Figure 5b, black lines). Transgenic T cells did not proliferate when incubated with DCs exogenously loaded with βgal peptide or DCs from mice vaccinated with βgal-LdTM-infected insect cells (Figure 5b, dashed lines). These results indicate that peptides from infected insect cells may be cross-presented by DCs in the spleen within 24 h of vaccination.
We previously showed that vaccination with peptide 39 protects mice from tumor challenge (27). To ensure that vaccination with peptides produced in insect cells elicits similar antitumor responses as synthetic peptides, we tested the vaccine in tumor protection and therapeutic assays. Mice were vaccinated with 39-LdTM-, AH1-LdTM-, or βgal-LdTM-infected insect cells 14 and 7 days prior to subcutaneous challenge with 5×104 CT26 tumor cells. The timing of this tumor challenge correlates with the peak of the expansion of AH1-Ld tet+ T cells 14 days after the initial vaccination (Figure 2b). Tumor growth was monitored for 60 days by palpation of the injection site. As indicated on the Kaplan-Meier plot, mice were sacrificed when their tumors reached 100 mm2. Vaccination with 39-LdTM-infected insect cells protected the majority of mice from subsequent CT26 tumor challenge, while vaccination with βgal-LdTM-infected, AH1-LdTM-infected, or uninfected insect cells failed to protect against tumor development (Figure 6a and data not shown). As expected, vaccination with the high affinity mimotope 15 protected significantly fewer mice from tumor formation than the intermediate affinity mimotope 39. The response to infected insect cells depends on the presence of CD8+ T cells, as CD8 antibody depletion of mice vaccinated with 39-LdTM-infected insect cells results in tumor growth in all mice tested (Figure 6a).
We next determined the therapeutic efficacy of vaccination with infected insect cells. We injected mice with 5×104 tumor cells and vaccinated 2, 4, 7, 10, and 14 days later with 39-LdTM-, βgal-LdTM-infected, or uninfected insect cells (Figure 6b). We observed a statistically significant delay in tumor growth in mice injected with 39-LdTM-infected insect cells, although no mice remained tumor free. These results show that antigen-specific T cells, elicited by the infected insect cell vaccine, slowed the growth of the tumor but did not eliminate it.
Finally, we determined if peptides identified in the BV peptide-library could be analyzed in vivo using this method. The synthetic WMF peptide [SPTYAYWMF (23)], a mimotope of the AH1 antigen, was identified in a BV-peptide library with substitutions in the MHC-contact residues. This peptide is insoluble in water and is difficult to synthesize indicated by the heterogeneity of HPLC and mass spectrometry profiles (data not shown). However, the WMF peptide produced in infected insect cells binds to CT-TCR with high affinity relative to peptide 39 (Figure 7a) and stimulates a corresponding amount of proliferation of a T cell clone expressing the CT-TCR (Figure 7b). These experiments demonstrate that the WMF peptide-H-2Ld complex is produced in insect cells and specifically binds to both soluble CT-TCR molecules and to T cells expressing CT-TCR.
We next vaccinated mice with insect cells infected with BV encoding WMF-LdTM to analyze the tumor-specific T cell responses and to determine if antitumor activity is afforded by this high-affinity mimotope. Vaccination with 39-LdTM and WMF-LdTM-infected insect cells elicited tumor antigen-specific T cells, as determined by tetramer staining (Figure 7c). Although the affinity of the WMF peptide in the peptide-MHC/TCR interaction is higher than the intermediate affinity 39 peptide, fewer AH1-specific T cells were detected in the blood after vaccination with WMF-LdTM infected insect cells. In addition, vaccination with 39-LdTM-infected insect cells protected the majority of mice from subsequent CT26 tumor challenge, while vaccination with βgal-LdTM, AH1-LdTM, or WMF-LdTM-infected insect cells failed to protect (Figure 7d). These results are consistent with those in Figure 6 and our previous results showing that peptides that form an intermediate affinity peptide-MHC/TCR complex with substitutions in the MHC-contact residues, such as the 39 peptide, protect mice from tumor challenge (27), while those that form a low (AH1) or high (peptide 15 or WMF) affinity peptide-MHC/TCR complex do not.
We have shown that vaccination with BV-infected insect cells expressing peptide-MHC complexes generates functional antigen-specific CD8+ T cell responses. This vaccine, which protects against tumor challenge and delays growth of tumors, is convenient in that it does not require additional adjuvants to those provided by the BV and insect cells, it eliminates the need, expense, and potential artifacts of synthesizing peptides identified in the baculovirus peptide-library prior to evaluation, and it does not require purification of proteins or BV prior to injection. Vaccination with infected insect cells expressing other recombinant proteins also results in specific immune cell responses. For example, we have used this vaccination strategy to develop TCR-specific antibodies (data not shown). Like with the peptide-MHC vaccine, vaccination with infected insect cells expressing recombinant proteins eliminates the need for protein purification.
This vaccine strategy is unique and cannot be practically achieved using other strategies because it provides a method to analyze peptides identified in BV peptide-MHC libraries that are otherwise technically difficult to evaluate, such as the WMF peptide (Figure 7). Amino acids such as cysteine, methionine, and tryptophan are often avoided in peptide libraries due to disulfide bonding, insolubility, and sensitivity to oxidation (27, 42). Small molecular changes in amino acid residues of the peptide, such as oxidation or alkyl-chain modifications, can alter epitopes and thus elicit different T cell responses (43, 44). Furthermore, vaccination with synthetic peptides does not always stimulate T cells that recognize endogenously processed peptides, possibly due to oxidation or cysteinylation of amino acid residues during peptides synthesis or handling (45–47). Although oxidation does not affect the T cell response to all antigens, in the BV-infected insect cell vaccine strategy, the peptide is produced, processed, and cross-presented in a reduced intracellular environment similar to natural tumor antigens.
In vitro characterization of peptides derived from libraries requires co-expression of MHC molecules and peptides. Insect cells infected with recombinant BV encoding peptide, H-2Ld, and β2m molecules bind H-2Ld-specific antibodies and the AH1-specific CT-TCR, suggesting that the protein structure is similar to that of mammalian cells. Furthermore, the avidity of the peptide-MHC/TCR complexes correlates with the affinity of the soluble mimotope-MHC/TCR complexes (Figure 1) (27). In addition to binding assays for characterization of peptides, other in vitro studies using infected insect cells examine T cell function, such as cytokine production (21, 22). In vivo, insect cells expressing peptide 39 restricted by H-2Ld elicit a population of T cells that binds antigen-loaded tetramer (Figure 2a and b), kills antigen-loaded target cells (Figure 2c and d), and protects 67% of mice from subsequent tumor challenge (Figure 6a). These proof-of-concept experiments indicate that the T cells elicited by infected insect cells recognize native peptide on tumor cells.
To analyze the mechanism of priming by this vaccine, we developed a new transgenic mouse that expresses the α and β chains of a TCR specific for the AH1/H-2Ld antigen. This TCR was derived from a BALB/c mouse vaccinated with irradiated CT26 tumor cells expressing GM-CSF (28); i.e., a T cell clone that had escaped negative selection in the thymus. Since T cells from this mouse recognize a tumor/self antigen, we are including them in our analyses to determine the requirements of peptide vaccines that break tolerance. Not all tumor-specific T cells generated in this mouse express CD8 molecules, suggesting that down regulation of the CD8 molecule is a consequence of tolerance, as reported by others (48, 49). Alternatively, aberrant expression of the TCR during T cell development may disrupt the expression of the CD8 molecule. The antigen-specific response of the CD3+ CD8− T cells is less robust relative to the CD3+ CD8+ cells, suggesting that the co-receptor contributes to the binding avidity of the TCR complex. Consistent with this possibility, more co-receptor negative T cells bind 39-Ld tet than AH1-Ld tet (data not shown).
Although the BV constructions require the MHC molecules for peptide studies in vitro, it is not required to elicit specific T cells in vivo. Like other effective antitumor CD8+ T cell responses (41, 50), the results we show here are consistent with tumor-specific T cells elicited by cross priming. 1) Direct recognition of antigen on the surface of infected insect cells is not required to elicit T cells when the vaccine encodes peptide and MHC (Figure 4). Insect cells infected with BV encoding peptide-Ld complexes that are not expressed on the cell surface elicit a similar frequency of AH1-Ld tet+ T cells as insect cells encoding surface peptide-Ld complexes (Figure 4a). 2) When the vaccine encodes peptide, but no MHC molecules (peptide-β2m), similar responses are elicited (Figure 4). In this experiment the only MHC available to present peptide is from the host cells, not the vaccine. 3) This vaccine induces maturation of CD11c+ dendritic cells from the draining lymph nodes (mesenteric) and spleen as determined by increased expression of costimulatory and maturation markers CD80, CD86, MHCII, and CD70 (Figure 5a). The expression of CD70, a maturation marker whose expression correlates with optimal expansion of CD8+ T cells following vaccination with both CD40 ligands and TLR agonists (51), increased on DCs after vaccination. 4) CD11c+ cells from vaccinated mice stimulate transgenic T cells to proliferate in an antigen-specific manner ex vivo. No additional antigen or adjuvant is added in these experiments (Figure 5b). 5) When the vaccine expresses costimulatory molecules (ICAM and B7) the T cell response to the vaccine is unchanged (data not shown). 6) Finally, it is unlikely that peptide-MHC molecules are produced in BV-infected dendritic cells because the polyhedron promoter driving transcription of the antigens are active only in insect cells (9–11). This feature of BV makes them safe to work with. Alternatively, extracellular antigen processing and presentation by dendritic cells stimulates CD4+ T cells (52). A similar mechanism in which MHC I-restricted peptides are loaded onto dendritic cells is possible. However, because this vaccine delivers only an estimated 200 ng of peptide and 10 µg of free peptide are required for similar responses (27), this mechanism likely accounts for a small fraction of the T cell response. Thus, like other vaccine delivery systems (53), the injected insect cells do not present antigens directly, but activate T cells by transferring the antigen to host professional APCs resulting in effective priming of tumor-specific T cell responses. In summary, use of these BV peptide-MHC constructions provides a streamlined system for evaluation of newly discovered peptides.
We would like to thank Drs. Denise Golgher, Haruo Tsuchiya, and Su-Yi Tseng for assistance with the production of the CT-TCR transgenic mice. We would also like to thank Jennifer McWilliams for constructing the CT-TCR plasmid, Heather Knowles for assistance with the immunoprecipitation, and Dr. Phillip Sanchez for help with the dendritic cell activation experiments.
1This work was supported by NCI CA109560 and a seed grant from the ACS IRG/UCCC to J.E.S. K.R.J., R.H.M., and J.Z.O. were supported in part by the Cancer Research Institute Predoctoral Emphasis Pathway in Tumor Immunology Fellowship.