DcR3/TR6, a secreted protein belonging to the TNF receptor superfamily, interacts with lymphotoxin-like, exhibits inducible expression, and competes with herpes simplex virus glycoprotein D for herpes virus entrance mediator (LIGHT), Fas ligand (FasL), and TL1A, all members of the TNF superfamily. Solid-phase TR6 can trigger reverse signaling of LIGHT and FasL expressed on T cells, and lead to T cell costimulation. In this study, we engineered tumor cells to express cell surface TR6 and used these cells as a tumor vaccine. We demonstrated that mastocytoma P815 cells expressing surface TR6 (TR6-P815) effectively augmented the T cells response in vitro and ex vivo in terms of proliferation, as well as IL-2 and IFN-γ secretion. TR6-P815 cells had reduced tumorigenicity compared with parental P815 cells. When inactivated TR6-P815 cells were employed as a vaccine, they protected the mice from challenge with live parental P815 cells, and eliminated established P815 tumors. The cell surface TR6-based tumor vaccine was also effective against low antigenicity tumors, such as B16 melanoma; co-administration of bacillus Calmette-Guérin further enhanced the vaccine’s efficacy. Thus, cell surface TR6 expression is a useful addition to our tumor vaccine arsenal.
DcR3/TR6 belongs to the TNF receptor superfamily (1, 2, 3). It is a secreted protein due to the lack of transmembrane domain in its coding sequence (3). TR6 has three ligands, i.e., Fas ligand (FasL), 4 lymphotoxin-like, exhibits inducible expression, and competes with herpes simplex virus glyco-protein D for herpes virus entrance mediator (LIGHT), and TL1A. As a secreted protein, it does not transduce signals into cells, although it is a member of the TNF receptor family, but is capable of interfering with the function of its ligands: TR6 can inhibit FasL-, LIGHT- and TL1A-induced apoptosis (3, 4, 5), and LIGHT-triggered T cell costimulation via a LIGHT receptor, HveA.
LIGHT, which belongs to the TNF superfamily, binds to three members of the TNF receptor family, i.e., HveA (6, 7), LTβR (8, 9), and TR6 (3). In the immune system, LIGHT is expressed on activated T lymphocytes, NK cells (10, 11), and immature dendritic cells (12). We have demonstrated that resting T cells also express a considerable amount of LIGHT on their surface, but it is better detected by confocal microscopy than by flow cytometry (13). LIGHT can induce apoptosis in cells expressing both HveA and LTβR (3) or LTβR alone (14). Recent studies have shown that LIGHT can costimulate T cell responses via its receptor HveA in vitro and in vivo (15). Transgenic mice overexpressing LIGHT have augmented immune responses (16), and LIGHT knockout (KO) mice present defects in cytotoxic T cell activity (15, 17, 18). Taken together, these lines of evidence indicate that LIGHT acts on HveA for T cell costimulation.
Certain cell surface ligands can receive stimuli from their receptors and transduce signals into the cell. Such a phenomenon is termed “reverse signaling”, because in this case the ligands function as receptors, whereas the receptors function as ligands. Some members of the TNF superfamily are capable of reverse signaling. Lanier and colleagues (19) and Gray and colleagues (20) have demonstrated that CD40L transduces costimulation signals into T cells. Wiley et al. (21) reported that CD30L cross-linking activates neutrophils, and Cerutti et al. (22) showed that such reverse signaling inhibits Ig class switch in B cells. Reverse signaling through membrane TNF-α confers resistance of monocytes and macrophages to LPS (23). Cross-linking of TNF-related activation-induced cytokine (TRANCE) enhances IFN-γ secretion by activated Th1 cells (24). Reverse signaling through FasL can promote maximal proliferation of CD8 CTLs (25, 26, 27). Cross-linking of TNF-related apotosis-inducing ligand (TRAIL) by its solid-phase death receptor 4 increases IFN-γ production and T cell proliferation (28). In the case of CD40L, its ligation results in general protein tyrosine phosphorylation, Ca2+ influx, and Lck, PKC, JNK, and p38 MAPK activation in EL4 thyoma cells (29, 30). TRAIL cross-linking also induces p38 MAPK activation (28). Our studies proved that LIGHT can reversely transduce signals into T cells when stimulated with solid-phase TR6, and such signaling can costimulate T cells (13, 31). With these new findings on LIGHT reverse signaling, the results from LIGHT transgenic and KO mice can be reinterpreted. Increased LIGHT reverse signaling might contribute to the augmented immune responses observed in LIGHT transgenic mice (16); conversely, elimination of such reverse signaling might contribute to the abated immune responses seen in LIGHT KO mice (15, 17, 18). Such reinterpretation does not refute the importance of forward LIGHT costimulation mediated by HveA.
Most tumors express unique Ags due to genetic alterations, or express tissue-specific Ags and/or developmental Ags owing to epigenetic effects. However, tumor cells have strategies to evade immune surveillance: down-modulating Ag processing and presentation to T cells; secreting soluble factors (e.g., cytokines and decoy receptors) to dampen or deviate immune responses; inducing T cell tolerance through multiple pathways (32, 33, 34, 35, 36, 37, 38). Several experimental approaches have been employed to enhance the immunogenicity of tumors and to break T cell tolerance in tumor-bearing animals or humans (38, 39). One such approach is to express, on tumor cells, costimulatory molecules, such as B7–1 (40, 41), B7H (42), 4–1BBL (43), or anti-4–1BB Ig H chain (44). Costimulatory molecules have also been coexpressed on tumor cells with other immune-enhancing factors, for example, B7–1 plus IL-2 (45), B7–1 plus CD2 ligand (46), B7–1 plus B7–2 plus 4–1BB (47); this strategy can even work for tumors with low antigenicity (43, 48, 49, 50). Animals inoculated with such manipulated tumor cells show retarded tumor growth and prolonged survival, compared with those receiving wild-type tumor cells (40, 42, 43, 51, 52, 53). When such tumor cells are used as vaccines, they can generally protect mice from the challenge of parental wild-type tumors. In a limited number of cases, when animals are vaccinated with tumors expressing anti-4–1BB mAb single chain (44), or inoculated with double recombinant adenovirus expressing B7–1 and IL-2 in the tumor mass (45), the existing tumors manifest significant regression. These findings imply that although therapeutic vaccination might not be effective with all costimulating molecules, it does have in certain cases therapeutic effects. This is important in clinical situations, in which tumor burden inevitably already exists.
We have reasoned that it is advantageous to use membrane-bound TR6 to enhance T cell costimulation and augment tumor antigenicity because, in theory, solid-phase TR6 can act on at least two reverse signaling pathways, i.e., via LIGHT and FasL, to achieve T cell costimulation. In this study, we engineered the surface expression of normally soluble TR6 on tumor cells and explored their usefulness as a tumor vaccine.
Materials and Methods
Mice and cell lines
Six- to 8-week-old female DBA/2, C57BL/6, or nude (nu/nu CD1) mice were purchased from Charles River Canada. The murine mastocytoma line P815 and the melanoma line B16-F10 were obtained from American Type Culture Collection. The human 293 primary embryonic kidney cells were procured from Qbiogene. The cell lines were cultured in complete DMEM medium containing 10% FCS, l-glutamine, and antibiotics.
Plasmid construct for cell surface TR6 expression
The coding sequence of human TR6 minus the stop codon was retrieved by PCR and fused with the coding sequence of the human EphB6 transmembrane domain (E591-R621) (54) followed by a stop codon; the ensemble was cloned into pAdenoVator (Qbiogene) downstream of the CMV promoter. The insert and junctions were verified by sequencing. The construct was named pCMV-TR6Mem.
Stable transfection of tumor cells with pCMV-TR6Mem
pCMV-TR6Mem and pcDNA III were introduced into P815 cells, 293 cells and B16 cells at a 10:1 ratio (20 μg of pCMV-TR6Mem and 2 μg of pcDNA per 5 × 106 cells) with electroporation (350 V, 960 μF). The transfected cells were then selected with 800 μg/ml G418 for 2 wk. The stable transfectants, named TR6-P815, TR6–293 and TR6-B16, were used for experiments after cell surface expression of TR6 was confirmed by flow cytometry. The parental cells were transfected with the empty vector pAdenoVator, and the resulting cells were named vector-293, vector-P815, and vector-B16.
Mouse and human T cell isolation and culture
Cells were flushed from the mouse spleen, and RBC were lysed with 0.84% NH4Cl, as described elsewhere (13). The resulting cells were referred to as spleen cells. Human PBMC were prepared with density gradients using Lympholyte-H (Cedarlane). Mouse and human T cells were purified from spleen cells and PBMC, respectively, with human and mouse T cell purification columns according to the manufacturer‘s instructions (Cedarlane). In some experiments, mouse T cells were further fractionated into CD4 and CD8 cells by magnetic bead positive selection (Miltenyi Biotech) according to manufacturer’s instructions. The purity of the CD4 and CD8 cells was ∼97% according to flow cytometry. In other experiments, CD4 or CD8 cells were deleted from spleen T cells with the magnetic beads. The cells were cultured in RPMI 1640 supplemented with 10% FCS, l-glutamine, and antibiotics. [3H]Thymidine uptake was measured as described previously (13, 55).
IL-2, IL-4, and IFN-γ in culture supernatants were measured with ELISA kits from R&D Systems according to the manufacturer’s instructions.
The binding of TR6-Fc to LIGHT−/− or LIGHT+/+ T cells was demonstrated using confocal microscopy, as detailed in our previous publication (13). Briefly, spleen cells were cross-linked with anti-CD3 and anti-CD28 for 2 min, and the cells were stained with Alexa-594-anti-CD3 for the TCR complex, and stained with TR6-Fc followed by Alexa-488-anti-human IgG for TR6 binding. Three view fields from confocal microscopy were randomly selected, and TR6-binding cells in green and TCR-positive cells in red were counted. The percentage of TR6+ cell among CD3+ cells was calculated.
Cell surface expression of TR6 on pCMV-TR6Mem-transfected cells was assessed by flow cytometry. The cells were stained with rabbit anti-TR6 polyclonal Ab followed by PE-conjugated goat anti-rabbit Ab (Cedarlane). Propidine iodine-negative live cells were gated for analysis. Vector-transfected cells served as controls.
DBA/2 mice (H-2d) were immunized s.c. with 1 × 106 mitomycin C-treated TR6-P815 cells (H-2d) on days 1 and 7. On day 15, spleen cells from these immunized mice were stimulated with mitomycin C-treated wild-type P815 cells at a 1:1 ratio in 24-well plates (8 × 106/4 ml/well). CTL activity of the cultured cells was assessed 6 days later by standard 51Cr release assay, as described previously (13), using 51Cr-labeled P815 cells (H-2d) as targets at different effector/target ratios. The lysis percentage of the test sample was calculated as follows: % lysis = (cpm of the test sample − cpm of spontaneous release)/(cpm of maximal release − cpm of spontaneous release).
Tumorigenicity and tumor challenge assays
For tumorigenicity assay, DBA/2 or nude mice (nu/nu CD1) were inoculated s.c. with 5 × 104 syngeneic parental P815 cells. Tumor size was measured once every 2 days (q2d) until day 30 after inoculation. For tumor challenge assays, DBA/2 mice were vaccinated twice with 1 × 106 mitomycin C-treated TR6-P815 cells on days 1 and 7. On day 15, the mice were challenged with 5 × 104 syngeneic parental P815 cells or 5 × 105 SP A/20 myeloma cells (H-2d). The product of the longest and shortest tumor diameters was taken as tumor size in this and all other experiments.
DBA/2 or C57BL/6 (H-2b) mice were first inoculated with 5 × 104 wild type P815 cells or 1 × 105 B16 (H-2b) cells, respectively. On days 3 and 8, the mice were vaccinated with 5 × 106 mitomycin C-treated TR6-P815 or TR6-B16 cells.
Cell surface expression of TR6
The coding sequence of human TR6 lacks the transmembrane domain, and consequently, TR6 is a secreted protein. To allow TR6 to anchor on the cell surface, we fused the coding sequence of the human EphB6 transmembrane domain (E591-R621) to the 3′ end of the TR6 coding sequence (Fig. 1 A). The fused sequences were cloned into pAdenoVator downstream of the CMV promoter, and the resulting construct was named pCMV-TR6Mem. This construct was stably transfected into human 293 embryonic kidney cells, high antigenic P815 mouse mastocytoma cells, and low antigenic mouse B16 melanoma cells.
Surface TR6 expression of the transfectants was verified by flow cytometry. As shown in Fig. 1 B, 55.8% of 293 cells, 70.3% of P815 cells, and 68.7% of B16 cells became cell surface TR6-positive after stable transfection, whereas the wild-type cells or cells transfected with control vectors remained TR6 negative. These pools of transfected cells without further cloning (to avoid selecting clones with different antigenicity) were used for in vitro or in vivo studies.
T cell costimulation by cell surface TR6 in vitro
We have previously demonstrated that when TR6 was placed on the solid phase of culture wells, it was able to costimulate T cells activated with a suboptimal concentration of anti-CD3. To assess whether this feature of TR6 could serve as a tumor vaccine, we first tested in vitro whether cell surface TR6 could similarly costimulate T cells. When human T cells were stimulated with mitomycin C-treated parental or control vector-transfected allogeneic 293 kidney cells, little proliferation was detected from days 3 to 5 (Fig. 2,A, left panel), indicating that 293 cells are not effective in costimulation, even though the Ag on 293 cells was alloantigenic with respect to PBMC. When TR6–293 cells were used as stimulators, they induced drastic T cell proliferation; this shows that cell surface TR6 expression provides T cells with potent costimulation. Similar findings were made with mouse P815 cells. In the presence of a suboptimal concentration of soluble anti-CD3, T cells from DBA/2 mice failed to proliferate in response to stimulation from syngeneic parental P815 or vector-P815 cells (Fig. 2 A, right panel). When TR6-P815 cells were tested as stimulators, they triggered significant proliferation from days 3 to 5, suggesting that the cell surface TR6 acts as an effective costimulating molecule.
In Fig. 2 B, we demonstrated in the same experiment that TR6-Fc coated on wells or expressed on cell surface could both costimulate T cell proliferation. In the left panel, T cell proliferation was enhanced by TR6 on wells in the presence of suboptimal anti-CD3 (0.8 μg/ml for coating), whereas anti-CD3, TR6-Fc, or normal human IgG (used as a control for TR6-Fc) alone had minimal effect. In the right panel, surface TR6 expressed by P815 cells in the presence of soluble anti-CD3 (5 ng/ml, suboptimal) was shown to costimulate T cells as well, whereas soluble anti-CD3 at this concentration, wild-type P815 plus anti-CD3, or vector-transfected P815 plus anti-CD3 failed to do so. It is to be noted that the magnitude of T cell proliferation costimulated by TR6 on wells and by TR6 on cell surface does not reflect the potency of these two ways of costimulation, as the concentration and the route of CD3 administration, which are major factors in determining the strength of proliferation, in these two systems were totally different.
We further showed that highly purified CD4 and CD8 cells similarly responded to TR6-P815 costimulation (Fig. 2 C) in terms of proliferation. This indicates that cell surface TR6 directly costimulates CD4 and CD8 cells without involvement of non-T cells.
Culture supernatants from the above-described experiments were harvested on days 2, 3, and 4 for lymphokine assay. Responding to parental 293 or vector-293 cells, human T cells secreted negligible amounts of IFN-γ, IL-2, and IL-4; they produced high levels of IFN-γ and IL-2 but not IL-4 when stimulated with TR6–293 cells (Fig. 2,D). The lymphokine production of mouse T cells responding to P815 cells was very similar to that of human T cells: TR6-P815 cells were able to stimulate IFN-γ and IL-2 but not IL-4 secretion, whereas parental and vector-transfected P815 cells were not effective (Fig. 2 D). These findings corroborate the T cell proliferation data, and suggest that cell surface TR6 is capable of costimulating T cells and inducing Th1-like lymphokine secretion. It is worth mentioning that our ELISA was sufficiently sensitive to detect IL-4 produced by T cells after anti-CD3 and anti-CD28 stimulation (data not shown), indicating that the failed detection of IL-4 in TR6–293- or TR6-P815-stimulated T cells was not due to the lack of assay sensitivity.
We next investigated the nature of cell surface ligand with which TR6 interacted. We showed that soluble human LIGHT (10 μg/ml) effectively blocked TR6-P815-triggered costimulation in this system (Fig. 2,E). Moreover, using spleen cells from LIGHT gene KO mice, we demonstrated that TR6 bound to 83% wild-type T cells (Fig. 2 F), but only ∼18% LIGHT−/− T cells. Based on these results along with our previous studies related to the ligand of TR6 on T cells (13, 31), it seems that LIGHT on the T cell surface is, at least, an important signal recipient of the cell surface TR6.
Spleen T cells from mice immunized with TR6-P815 cells presented an augmented response to in vitro restimulation
Next, we assessed whether surface TR6 could augment the in vivo T cell response to tumor cells. DBA/2 mice were vaccinated with syngeneic mitomycin C-treated TR6-P815 cells on days 1 and 7. Parental and vector-transfected P815 cells served as controls. On day 15, the spleen cells of the vaccinated mice were harvested and restimulated with mitomycin C-treated parental P815 cells in vitro. The cell proliferation was measured by [3H]thymidine uptake on days 4, 5, and 6, lymphokine production of the culture was assayed on days 3, 4, and 5 by ELISA, and CTL activity was quantified on day 6 by 51Cr-release assay. As shown in Fig. 3,A, parental and vector-P815 cells failed to prime the immune system in vivo, as the spleen T cells from these mice did not respond to secondary in vitro restimulation with parental P815 cells in terms of proliferation (Fig. 3,A), IFN-γ and IL-2 production (Fig. 3,B), and CTL activity (Fig. 3,C). However, when the mice were vaccinated with TR6-P815 cells, their spleen T cells became highly responsive to in vitro restimulation by parental P815 cells: T cells were strongly proliferative on days 4 and 5 (Fig. 3,A); IFN-γ and IL-2 but not IL-4 were detected at high levels in the culture supernatants on days 3 and 4 (Fig. 3,B); they also showed augmented CTL activity against parental P815 cells when stimulated wild-type P815 (Fig. 3,C). We further demonstrated that both CD4 and CD8 cells were essential during restimulation, because spleen cells from the immunized mice had significantly reduced CTL against wild-type P815 cells, if CD4 or CD8 cells were depleted (Fig. 3D). These results revealed that TR6 expression on the tumor cell surface converts the normally ineffective immune response against the tumor cells into an effective one in vivo, and the secondary immune response depends on both CD4 and CD8 cells. This has established the basis of using tumor cells with surface TR6 expression as a tumor vaccine.
Reduced tumorigenicity of P815 cells after surface expression of TR6
Since P815 cells with surface TR6 expression could effectively trigger an in vivo T cell response, they should have a reduced capability of evading the immune surveillance and, hence, decreased tumorigenicity. This possibility was investigated here. First, different numbers of parental P815 cells were inoculated s.c. into syngeneic DBA/2 mice to determine the minimal tumorigenic number (MTN) required to achieve 100% solid tumor occurrence at the inoculation site. The MTN for P815 was determined to be 1 × 104 cells, and 5 × MTN (5 × 104) was used in all subsequent experiments. As shown in Fig. 4 A, all DBA/2 mice inoculated with the 5 × MTN of parental or vector-P815 cells developed solid tumors at the injection site, and tumor size reached 400 mm2 within 20 days. The mice were sacrificed at that time according to Canadian Council on Animal Care guidelines. However, none of the mice inoculated with TR6-P815 cells developed solid tumors at the injection site during the observation period (30 days), and they had no visible tumors in their internal organs upon necropsy on day 30. Differences between the parental P815 vs the TR6-P815 group, and between the vector-P815 vs the TR6-P815 group were both highly significant (p < 0.001, one-way ANOVA followed by all pair-wise multiple comparison procedures (Tukey test)). Interestingly, three of six mice inoculated with TR6-P815 cells had small (<20 mm2) transient tumors at the injection site between days 12 and 15, but these tumors disappeared afterwards; the kinetics of tumor disappearance coincided with that of an effective anti-tumor immune response.
The failed formation of solid tumors by TR6-P815 cells in DBA/2 mice was not due to the reduced growth rate of these cells after stable transfection with pCMV-TR6Mem, because these cells had a similar in vitro proliferation rate compared with parental or vector-P815 cells (Fig. 4,B); moreover, in vivo, when the wild-type P815, vector-P815, and TR6-P815 were injected into T cell-deficient nude mice, they all formed tumors at rates with no statistical difference (Fig. 4 C). This experiment also demonstrates that the protective effect of cell surface TR6-based vaccine depends on host T cells, because vaccine was not effective in the absence of T cells in the nude mice.
TR6-P815 vaccination prevented tumor development on subsequent parental P815 cell inoculation and protection was tumor-specific
The failure of solid tumor formation after live TR6-P815 cell inoculation suggested that the recipient mice developed an effective immune response against the tumor and subsequently eliminated the inoculants. If such is the case, inactivated TR6-P815 cells could be used as vaccine to counter the challenge by live wild-type P815 cell inoculation. This possibility was therefore explored. When live wild-type P815 cells were inoculated into mice previously vaccinated with mitomycin C-inactivated wild-type P815 cells or vector-P815 cells, all of them developed tumors at the injection site after 14 days, and the tumors reached 400 mm2 around day 24 (Fig. 5 A). In contrast, none of the TR6-P815-vaccinated mice developed tumors upon the live parental P815 cell challenge. The differences between the parental P815- vs the TR6-P815-vaccinated groups and between vector-P815- vs the TR6-P815-vaccinated groups were highly significant (p < 0.001, one-way ANOVA followed by all pair-wise multiple comparison (Tukey test)). This indicates that mice vaccinated with inactivated TR6-P815 cells, but not parental P815 or vector-P815 cells, mounted an effective secondary anti-tumor immune response, which eliminated subsequently inoculated live parental P815 tumor cells.
Is this anti-tumor response specific? To answer this question, DBA/2 mice were inoculated with 5 × MTN of syngeneic SP A/20 myeloma cells (5 × 105 cells/mouse). These cells rapidly developed into solid tumors in the injection site in 7 days, and by day 14, the tumors reached 400 mm2 in naive DBA/2 mice as well as in DBA/2 mice vaccinated with TR6-P815 cells (Fig. 5B). This demonstrates that the anti-tumor immune response by TR6-P815 vaccination is tumor-specific.
TR6-P815 vaccine was effective in eliminating pre-existing P815 tumors
In most clinical cases, tumor vaccines will be administered to patients already diagnosed with tumors. Therefore, a tumor vaccine will only be realistically useful if it can eliminate pre-existing tumors. To this end, we inoculated DBA/2 mice with live parental P815 cells s.c. and administered tumor vaccination twice on days 3 and 8 in the opposite flank. As shown in Fig. 6, A and B, all mice vaccinated with parental P815 or vector-P815 cells developed tumors, and the kinetics were similar to those seen in mice without vaccination, as illustrated in Fig. 3,A. However, tumor development in 7 of the 10 mice vaccinated with TR6-P815 cells was totally prevented; 3 mice in this group did develop tumors, but tumor development was delayed for ∼4 days (Fig. 6 C). The differences between the parental P815- vs the TR6-P815-vaccinated groups and between the vector-P815- vs the TR6-P815-vaccinated groups were highly significant (p < 0.001, one-way ANOVA followed by all pair-wise multiple comparison (Tukey test)). This result demonstrates that tumor cells engineered to express cell surface-anchored TR6 can be used as a therapeutic vaccine.
TR6-B16 vaccine in combination with adjuvant was effective in eliminating pre-existing low antigenic tumors
P815 mastocytoma is a highly antigenic tumor, whereas many tumors in humans are of low antigenicity. To explore the general utility of cell surface TR6 in tumor vaccines, we expressed TR6 on the cell surface of B16 (H-2b), a low antigenic melanoma line. The resulting line was named TR6-B16 and was tested alone or in combination with bacillus Calmette-Guérin (BCG) as a tumor vaccine against parental B16 tumors. C57BL/6 mice (H-2b) were inoculated s.c. with 5 × MTN (1 × 105 cells/mouse) of live parental B16 tumors on day 0. These mice were mock vaccinated with PBS (Fig. 7,A) or vaccinated with vector-B16 (Fig. 7,B) on days 3 and 8. The two groups had similar tumorigenic kinetics and tumor incidence: all the mice developed palpable solid tumor at the injection site around day 11, and none of the mice were tumor-free on day 17. In mice vaccinated with TR6-B16 cells, the tumorigenic kinetics of parental B16 cells were delayed (Fig. 7,C); this was demonstrated more clearly when the data were expressed as percentage of mice with tumors (Fig. 7,G) and as percentage of mice with tumors reaching 400 mm2 at different days after inoculation (Fig. 7,H). Statistical analysis showed that the differences were highly significant or significant (p = 0.001 between the mock vaccinated and the TR6-B16 groups; p = 0.024 between the vector-P815 and the TR6-P815 groups; one-way ANOVA followed by all pair-wise multiple comparison (Tukey test)). When BCG was administered in combination with TR6-B16 cells, the effect became more pronounced: 5 of 9 mice were tumor free in this group (Fig. 7,F; compared with 0 of 8 mice being tumor free in the group with mock vaccination, 0 of 8 being tumor free in the groups with vector-B16 vaccination, and 0 of 6 being tumor free in the group with B16 plus BCG vaccination (Fig. 7,E) (p < 0.001 between the groups vaccinated with PBS and with TR6-B16 + BCG; p = 0.006 between the groups vaccinated with vector-P815 and with TR6-P815 + BCG; p = 0.33 between the groups vaccinated with PBS and with P816 + BCG; one-way ANOVA followed by all pair-wise multiple comparison (Tukey test)). BCG is known to moderately enhance immune responses to certain types of tumors (56). However, the significant effect of tumor elimination in the TR6-B16 plus BCG-treated group could not be attributed to BCG alone, as the BCG alone-treated group had about only a 2-day delay in their tumor development (Fig. 7 D), compared with the mock vaccinated group, but none of the mice in this group was tumor free (n = 7). This indicates that TR6-B16 and BCG have additive or synergistic effect as a therapeutic tumor vaccine.
Low antigenicity and lack of costimulatory molecules are some of the reasons why tumor cells are invisible to immune surveillance. One of the approaches often taken to overcome these problems is to express costimulating molecules, usually from the Ig superfamily or TNF superfamily, on the tumor surface (57, 58, 59). We developed a novel strategy in this regard based on costimulation of T cells by the cell surface expression of TR6, which is a soluble protein of the TNF receptor superfamily.
In our previous study, we demonstrated that when TR6 is coated on wells, it enhances the T cell response to suboptimal stimulation of TCR. Those in vitro assays are subject to criticism that any molecule that can physically increase contact force between the culture well, on which anti-CD3 or anti-TCR is coated, and T cells has the potential to augment T cell responses; such a molecule might not fall into a more strict definition of costimulatory molecules. In this study, we proved that when TR6 was anchored on the tumor cell surface, it augmented T cell responses triggered by tumor Ag in vitro and in vivo. Therefore, TR6 on the cell surface can now be qualified as a genuine costimulatory molecule, albeit an artificial one, as it is not normally expressed on the cell surface.
Cell surface TR6-based tumor vaccine likely depended on adaptive immunity. First, in vitro study showed enhanced response of naive T cells to TR6–293 and TR6-P815 costimulation (Fig. 2); second, T cells from TR6-P815 vaccinated mice had augmented response to wild-type P815 cells, and such augmented response was reduced when CD4 or CD8 cells were deleted in vitro; third (Fig. 3), TR6-P815 vaccinated mice rejected parental P815 tumors but not a different tumor, i.e., SP A/20 myeloma cells (Fig. 5); finally, our data (not shown) showed that NK cell activity was not modulated by either solid-phase or soluble TR6.
The TR6 employed is of human origin, and mice do not have an orthologous counterpart of human TR6. Was the observed tumor immunity in mice due to the xenoantigenic nature of human TR6? Although we used TR6-P815 or TR6-B16 for immunization, the cells used for challenging or inoculation to generate solid tumors (e.g., in the P815 challenging assay (Fig. 5), in the P815 therapeutic assay (Fig. 6), and in the B16 therapeutic assay (Fig. 7)) were all wild-type tumor cells carrying no TR6, yet the tumor immunity could curb their growth in vivo, or eliminate them after the tumors were pre-inoculated. If one argues that the TR6 as xenoantigen enhanced the general nonspecific immune status, which led to tumor rejection, we have shown in Fig. 5 B that a difference tumor line SP A/20 was not affected by the vaccine. Therefore, the enhanced immune response is tumor-specific, but not related to the xenoantigenicity of TR6.
TR6 has three ligands, i.e., LIGHT, FasL, and TL1A, and the former two are capable of reverse signaling and costimulating T cells. According to our genome search, mice do not have an orthologous gene corresponding to human TR6. Therefore, we used human TR6 for both in vitro and in vivo study in a mouse model, because human TR6 can bind to mouse LIGHT and FasL (4, 60). We have shown that soluble LIGHT effectively blocks TR6-P815 triggered costimulation, and LIGHT−/− T cells had significantly reduced binding to TR6, compared with LIGHT+/+ T cells (a decrease from 83 to 18%). These results along with our previous study (13, 31) suggest an important role of LIGHT in the costimulation. With that said, we cannot exclude possible reverse signaling via FasL (25, 26, 27) or other so far unidentified TR6 ligands on T cells, because TR6 can still bind to ∼18% LIGHT−/− T cells according to confocal microscopy.
Our previous data showed that soluble and solid-phase TR6 decreases T cell chemotaxis (55). This is a possible mechanism to explain that tumors secreting soluble TR6 have less lymphocyte infiltration (61) because when T cells encounter soluble TR6 outside the tumor, they will decrease they migration toward the tumor. However, in our current model, TR6 is expressed on the tumor cell surface; T cells would not have chances to interact with the cell surface TR6 until they enter the tumor. Hence, the immune surveillance should not be hampered but possibly augmented by the cell surface TR6.
We have proved in this mouse model that tumor cells engineered to express surface TR6 could be used as a therapeutic vaccine. In humans, some patients with gastrointestinal tumors have elevated serum TR6; for some tumors, such as gastric, liver, gallbladder, and colon tumors, the serum TR6-positive incidence reaches 50–70% (61); serum TR6 is secreted by the tumors (2, 61). Conceivably, free serum TR6 might compete with cell surface TR6 in the cell surface TR6-based vaccine and reduce its efficacy. Therefore, such vaccine might be more effective in tumors that do not secrete soluble TR6. Our study shows that lung and breast cancers have a much lower incidence of serum TR6 positiveness (<20%) (61), and patients with these cancers should be optimal candidates for a surface TR6-based therapeutic tumor vaccine. Moreover, we have also shown that in tumors with high serum TR6 levels, tumor resection leads to complete disappearance serum TR6 in 3–4 wk; this will be the best time window for administration of surface TR6-based tumor vaccine, as tumor-specific T cell immunity can develop in the absence of TR6 interference in such a time window. In summary, tumor cells engineered to express surface TR6 are a useful addition in our arsenal of therapeutic tumor vaccines.
The authors have no financial conflict of interest.
The authors sincerely thank Ovid Da Silva and Robert Boileau for their editorial assistance and statistical analysis, respectively. The authors also thank Dr. Leiping Chen for providing LIGHT−/− mice to the Human Genome Sciences Inc.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by grants from the Canadian Institutes of Health Research (CIHR; MOP57697, MOP69089), the CIHR/Canadian Blood Service Partnership Program, the Kidney Foundation of Canada, the Heart and Stroke Foundation of Quebec, the Roche Organ Transplantation Research Foundation, Switzerland (ROTRF 590934439), Genome Canada/Quebec, and the J-Louis Levesque Foundation (to J.W.). Support was also received from the CIHR for the New Emerging Teams in Transplantation. J.W. is a National Scholar of Fonds de la recherche en santé du Québec.
Abbreviations used in this paper: FasL, Fas ligand; LIGHT, lymphotoxin-like, exhibits inducible expression, and competes with herpes simplex virus glycoprotein D for herpes virus entrance mediator; KO, knockout; TRANCE, TNF-related activation-induced cytokine; TRAIL, TNF-related apotosis-inducing ligand; MTN, minimal tumorigenic number; q2d, once every 2 days; BCG, bacillus Calmette-Guérin.