Immune-Mediated Regression of Established B16F10 Melanoma by Intratumoral Injection of Attenuated Toxoplasma gondii Protects against Rechallenge Free

Despite considerable progress using surgery, chemotherapy, and radiation to treat cancer, the 5-y survival rates for many cancers are still very low and not improving. There is currently a great deal of interest in developing therapies that stimulate effective immune responses against cancer to establish another major therapeutic option and improve outcomes. The immune system is stimulated by microorganisms, and since the studies of Coley (1) over 100 y ago, the possibility of using microorganisms as adjuvants to stimulate antitumor immunity has been recognized. However, despite frequent efficacy against what were deemed to be incurable, often metastatic cancers, “Coley’s toxins” were not accepted clinically. Since Coley’s time, we have developed a detailed understanding of both the immune system response to tumors and the suppression of the immune system by tumors. Genetic manipulation enables the generation of microorganisms with reduced virulence that can still function as powerful immunologic adjuvants, and this has led to progress in developing microorganisms as immune-stimulating antitumor reagents. For example, the standard of care for treating superficial bladder cancer is instilling Bacillus Calmette-Guerin into the bladder (2), and a variety of other microorganisms, such as Listeria monocytogenes, Salmonella typhimurium, and multiple viruses, are in various stages of development as antitumor vaccines and treatments (3–5). Importantly, although each of the most heavily studied organisms has efficacy in various models and some show promise in clinical trials, none of these organisms has been shown to eliminate an established, poorly immunogenic tumor.

Toxoplasma gondii is a single-cell, obligate intracellular, eukaryotic parasite. cps, the strain used in this study, is a uracil auxotroph due to deletion of the carbamoyl phosphate synthetase II enzyme (6). Because the required enzyme is inactivated, there is little potential for reversion. The cps grows well in vitro in mammalian cells in uracil-supplemented medium. In vivo, cps is nonreplicative, but efficiently invades cells and generates a strong adaptive immune response, characterized by activation of APCs to stimulate CD8⁺ T cell maturation and expansion with associated generation of high levels of IL-12 and IFN-γ (7). Because such immune responses are associated with effective antitumor immunity, we evaluated whether cps could induce the specific antitumor immunity required to eliminate an aggressive, notoriously difficult-to-treat tumor.

B16F10 melanoma has been used extensively as a poorly immunogenic, highly aggressive model for murine tumor immunotherapy studies (8). Shrinkage of established B16F10 has not been achieved with an immune-based monotherapy. Established B16 tumors have been treated at low frequency by combination immunotherapies, such as adoptive transfer of Ag-specific transgenic T cells along with TLR agonist administration (9), or at high frequencies by combining adoptive transfer of Ag-specific transgenic T cells and recombinant viral infection and preconditioning the host through total body irradiation (10). These approaches are complex and will be difficult to accomplish in a widespread clinical context. However, they do provide encouragement for cancer immunotherapy. New approaches that are easy to apply and have high levels of efficacy and minimal side effects need to be developed to improve cancer immunotherapy outcomes.

Our hypothesis was that cps introduced into the tumor microenvironment would transform that environment from one that is predominantly immunosuppressive to an immunostimulatory environment in which the endogenous antitumor immune response would effectively eliminate the tumors. In the studies reported in this work, we tested that hypothesis by using cps monotherapy to treat established B16F10 melanoma. We show that treatment of established B16F10 dermal melanoma by intratumoral injection of cps not only regresses the primary tumor, but also establishes systemic and memory antitumor immune responses with significant ability to reject or slow the development of rechallenge. The efficacy of cps monotherapy against the primary tumor and development of systemic response and immune memory, in combination with its inability to replicate in vivo and associated lack of observed toxicity, establishes the cps strain of T. gondii as a potentially powerful reagent for immunotherapy of solid tumors.

C57BL/6 were purchased from the National Cancer Institute. IL-12p35^−/− (002692), IFN-γ^−/− (002287), and NOD/SKID/IL-2Rγ^−/− (005557) mice were purchased from The Jackson Laboratory. Tumors were generated in Tyr-CreERt:BrafCA:Ptenlox/lox transgenic mice, on a mixed genetic background by intradermal (i.d.) injection of 4-hydroxy tamoxifen in the flank. Animal experiments were approved by the Institutional Animal Care and Use Committee at Dartmouth Medical School.

The B16F10 mouse melanoma cell line was originally obtained from I. Fidler (MD Anderson Cancer Center, Houston, TX) and passaged i.d. in C57BL/6 mice seven times to ensure reproducible growth. Tumor cells were cultured in RPMI 1640 containing 7.5% FBS and inoculated into mice only if viability exceeded 96%. A total of 1.25 × 10⁵ live B16F10 cells was inoculated i.d. in the right flank. Tumor diameters were measured three times weekly, and mice were euthanized when tumor diameters reached 15 mm. To determine T cell recall capacity, B16F10-bearing mice were treated as seen in Fig. 1A and then euthanized on day 20 of tumor challenge, and CD8⁺ T cell responses were assessed by flow cytometry or ELISPOT, as described below. The UpK10 tumor line was generated from C57BL/6 mice, as described (11). A total of 5 × 10⁵ live UpK10 cells was injected i.d. Lewis lung tumors were generated by injecting 1 × 10⁶ live cells, i.d. on the right flank.

Depleting anti-CD4 (clone GK1.5) and anti-CD8 (clone 2.43) were produced as bioreactor supernatants and administered i.p. in doses of 250 μg 1 day prior to treatment and then once weekly. Greater than 95% depletion of target T cell populations was confirmed by flow cytometry. Depleting anti-NK1.1 (clone PK136) was produced similarly and administered i.p. in doses of 200 μg for four treatments on days −2, 0, 2, and 9 relative to initial cps tumor treatment. CXCR3 blocking Ab (clone CXCR3-173) was purchased from BioLegend, administered i.p. in doses of 100 μg beginning 1 d before cps treatment, and continued every third day after for four total treatments. CXCR3 blockade was also performed at 200 μg, yielding similar results.

CD45 (30-F11), CD3ε (145-2C11), CD8b (YTS156.7.7), CD44 (IM7), CD11c (N418), and CD11b (M1/70) were purchased from BioLegend. NK1.1 (14B11) and IFN-γ (XMG1.2) were purchased from eBioscience. Fig. 2A–F flow cytometry samples were gated on CD45⁺ cells after the forward light scatter (FSC) and side light scatter (SSC) gate was set. Then Fig. 2B was gated on CD11c⁻CD11b⁺, whereas Fig. 2C was gated on CD11c⁺CD11b⁺ cell populations. Fig. 2D was gated on CD3⁺ cells, whereas Fig. 2E was gated on NK1.1⁺ populations and Fig. 2F was gated on Gr1⁺, CD11b⁻. The initial gating strategy for Fig. 2G was the same strategy used for Fig. 2A–F; however, a final gate was made that defined the population of cells infected. Infected cells were defined by the invaded presence of CSFE-stained cps parasites. Fig. 2H–L samples were first gated on CD45⁺ cells after the FSC and SSC gate was set. Fig. 2I was gated on CD3⁺CD8⁺ cell populations, and Fig. 2J was also gated on CD3⁺CD8⁺ cell, and then a histogram gate was made defining IFN-γ⁺ CD8⁺ cells. Fig. 2K was gated on NK1.1 cells, and Fig. 2L was first gated on NK1.1 and then gated on IFN-γ⁺ NK1.1.

Fig. 3A and 3B flow cytometry samples were first gated on CD45⁺ cells after the FSC and SSC gate was set. Then Fig. 3A was gated on CD3⁺CD8⁺ cells, followed by gating on the CD8⁺CD44⁺ cell population. Fig. 3B was gated on CD3⁺CD8⁻ cells, followed by gating on the CD8⁻CD44⁺ cell population.

Tumor explants were manually homogenized in T-Per Lysis Buffer (Thermo Scientific/Pierce Protein Research Products) with complete protease inhibitor mixture (Roche Applied Science), and total protein was quantified using Life Technologies/Invitrogen Qubit Protein Assay kit and an Invitrogen Qubit 2 fluorometer. IFN-γ, CXCL9, and CXCL10 protein were quantified using specific ELISA kits (R&D Systems) with SuperAquaBlue substrate (eBioscience); absorbance was measured using a BioTek Epoch microplate spectrophotometer. Cytokine and chemokine concentrations were calculated from a specific standard curve.

CD8⁺ T cells were isolated from spleens or draining inguinal lymph nodes and purified using anti-CD8 MACS magnetic beads (Miltenyi Biotec). CD8⁺ T cells were then plated at a 10:1 ratio with irradiated EL-4 thymoma cell targets (American Type Culture Collection) that had been pulsed with 1 μg/ml MHC-I–restricted peptide epitopes TRP-2 180–188, or OVA257–264, or heat-killed cps.

Flow cytometry was performed on a FACS-Canto system (BD Biosciences). Data were analyzed using FlowJo software (version 7.6).

Mice were i.d. injected with 1.25 × 10⁵ B16 cells and treated when tumor size reached 3.5–5 mm in diameter (unless otherwise stated). Tumors were injected four times with 1.5 × 10⁷ tachyzoites of cps suspended in PBS, as outlined in Fig. 1A. Mice were monitored daily and sacrificed when tumors reached 1.5 cm in diameter.

Cells from tumors were cultured for 5 h at 37°C in RPMI 1640 containing brefeldin A (10 μg/ml). Following incubation, cells were washed and stained with Abs against surface markers and then fixed, permeabilized, and stained intracellularly with anti–IFN-γ PE (clone XMG1.2; BioLegend). Flow cytometry was performed, as described above.

The cps was stained in 7.5 μM CFSE for 5 min at 37°C. After staining, cps was washed in PBS, and then 1.5 × 10⁷ tachyzoites were intratumorally injected into B16F10 tumor. Eighteen hours after second CSFE-labeled cps injection, cells were harvested from the tumor by surgically removing the tumor and disassociating with a scalpel, followed by flushing through a sterile cell strainer with 100-μm pores (catalog 22363549; Fisher Scientific). Strained cells were Ab stained using provider-suggested staining protocols.

The cps is identical to cps2-1 (6). T. gondii strains were maintained as tachyzoites by serial passage in human foreskin fibroblast cell monolayers. To grow cps, tissue culture medium was supplemented with 0.2 mM uracil. For mouse injections with T. gondii, parasites were purified by filtration through a 3.0-μm filter (Nuclepore; Steriltech, Kent, WA) and washed with medium or PBS.

All experiments were repeated at least twice with similar results between experiments. Survival experiments used between four and nine mice per group. Figures denote statistical significance of *p < 0.05, **p < 0.01, and ***p < 0.001. Statistical analysis was performed with GraphPad Prism 4 software. The p values <0.05 were considered significant. Survival experiments used log-rank Mantel Cox test for survival analysis. Data for bar graphs, including for the mean fluorescence of flow cytometry, are from a normal distribution applied to two independent groups, and the confidence interval was calculated using two-tailed unpaired Student t test. Error bars represent SEM from independent samples assayed within the represented experiments. ELISPOTs used pooled cells from multiple mice, and the values presented are from triplicate assay on the pools that were compared by one-way ANOVA.

B16F10 was inoculated i.d. into C57BL6 mice, and palpable 3.5- to 5-mm tumors (day 9–11 after inoculation) were treated with intratumoral cps injection (treatment schematic; Fig. 1A). The cps-treated tumors initially stopped growing and then rapidly regressed and were undetectable within 12 d of the first treatment in 100% of the treated mice (Fig. 1B, 1C). In comparison, by day 21, at which all tumors had disappeared in cps-treated mice, all mice treated with PBS had large tumors that constituted the experimental endpoint. The vast majority of the treated tumors (83%) did not return after the treatment (Fig. 1C), establishing, to our knowledge, cps treatment as the first immune-based monotherapy that can dependably regress established tumors of this very widely studied melanoma model.

Although cps does not express any tumor Ags, we also tested whether cps monotherapy could be therapeutic if injected into nontumor sites. The cps was administered to mice with 3.5- to 5-mm i.d. B16F10 both i.p. and i.v. Although intratumoral cps injection regressed the tumor, as expected, i.v. and i.p. cps injection did not slow tumor growth or improve survival in mice with established B16F10 melanoma (Supplemental Fig. 1). This demonstrates that, for the treatment of a primary tumor to be effective, cps must be in the tumor microenvironment, because i.v. or i.p. applied cps would not have accumulated in the tumor.

T. gondii latently infects between 10 and 70% of people in most countries (12). To determine how latent infection with T. gondii affects treatment efficacy, we infected mice with the Prugniaud (PRU) strain of T. gondii to establish a latent infection in mice (13) and tested the efficacy of treating dermal B16F10 with cps in these latently infected mice. Latent infection with PRU did not interfere with the ability of cps treatment to elicit an effective antitumor response (Fig. 1D). These data indicate that the widespread latent infection of humans by T. gondii is not likely to affect the efficacy of cps as a clinical antitumor treatment.

Tumors were treated with heat-killed cps to determine whether live cps is required for efficacy, or whether cps constituents act by direct stimulation of the innate immune response through recognition of molecular patterns of pathogens. Using the same treatment regimen associated with regression, there was a slight delay in tumor growth and a modest increase in survival time (Fig. 1E; p < 0.05). However, treatment with heat-killed cps failed to halt tumor growth or cause tumor regression in any mice, demonstrating that antitumor efficacy requires live organisms and suggesting that the organism must actively invade cells for efficacy.

A potential problem with live organism therapy is the possibility of establishment of active infection, especially in immunocompromised patients. To rigorously demonstrate that cps would be innocuous regardless of immune status, we administered the organism to NOD/SCID/IL-2R γ-chain knockout (KO) mice (NSG). These mice lack B, T, and NK cells (14). Intraperitoneal injection of 100,000 cps tachyzoites had no observable effect on immunocompromised mice (Supplemental Fig. 2), despite the fact that 100 tachyzoites of the parental RH strain of T. gondii are lethal to wild-type mice within 2 wk (15). These studies demonstrate that even in an immunocompromised host, the cps strain of T. gondii poses no safety risk as a treatment modality.

A previous study using i.p. injection of a nonattenuated strain of T. gondii demonstrated the ability to slow the growth of B16F10 and concluded that the effect was not adaptive immune system mediated, as tumor growth reduction could be achieved in immunodeficient mice (16). To conclusively prove that immune effector function is required for cps-mediated elimination of established B16F10 tumors, we evaluated cps treatment of B16F10 in severely immunodeficient NSG mice. The cps treatment failed to affect tumor growth in NSG mice (Fig. 1F), proving that cps-mediated tumor regression is mediated by the immune system.

Leukocyte distribution was assessed in untreated and cps-treated tumors by immunohistochemistry and flow cytometry. Both methodologies showed that cps treatment rapidly increased CD45⁺ leukocyte infiltration of the tumors by 2- to 3-fold after treatment (Fig. 2A). CD11b⁺CD11c⁻ macrophages were significantly increased in approximately the same proportions as total CD45⁺ cells (Fig. 2A, 2B). CD11c-expressing myeloid dendritic cells (DCs), NK1.1-expressing NK cells, and CD11b⁻Gr1⁺ neutrophils were not significantly changed as a proportion of CD45⁺ cells, although the increase in CD45⁺ cells means that the total cell numbers of these leukocytes did increase (Fig. 2C, 2E, 2F). Perhaps surprisingly, the percentage of CD3⁺ T cells in the tumor was significantly reduced in the early stages of the treatment. Overall this shows that the initial response to treatment is recruitment of myeloid, but not lymphoid cells.

To determine what cells are being invaded by cps, cps was labeled with CFSE, injected intratumorally on consecutive days, and assayed 18 h after the second cps injection. Flow cytometry revealed that ∼50% of CD45⁻ cells were CFSE positive and >80% of leukocytes were invaded (CD45⁺ cells; Fig. 2G). This demonstrates that leukocytes are first rapidly recruited to the tumor by cps injection (Fig. 2A) and preferentially invaded by cps. Macrophages, DCs, and NK cells were all 70–80% invaded, but only 20–30% of T cells and neutrophils were invaded (Fig. 2G). This shows that cps can invade a variety of nucleated cell types, but preferentially invades phagocytes, as is consistent with wild-type T. gondii.

To further investigate the kinetics of the adaptive immune response, we performed 2 d of cps injection, and then 3 d later analyzed the leukocyte populations in the tumor (Fig. 2H–L). Leukocytes (CD45⁺) were 2- to 3-fold higher in treated tumors (Fig. 2H, Supplemental Fig. 3), CD8⁺ T cells increased 3-fold (Fig. 2I), and 8% of these CD8⁺ T cells were producing IFN-γ as compared with 3% in the controls (Fig. 2J). Interestingly, whereas the fraction of NK cells was similar between treated tumors and the now much larger control tumors, there is a marked increase of NK cells producing IFN-γ in treated tumors (Fig. 2L). Together, these data show that cps treatment drives a rapid and durable leukocyte infiltration of the tumor (changes expressed in cell numbers rather than percentages are shown in Supplemental Fig. 4). At this later stage of treatment, there is an increase in CD8⁺ T cells and increased expression of IFN-γ by both CD8⁺ T cells and NK cells.

To examine systemic changes of immune cells, we evaluated changes in draining lymph nodes and spleens at day 20, 11 d after treatment was begun and a time at which treated tumors were all visually eliminated. We detected a statistically significant increase in Ag-experienced (CD44⁺) CD8⁺ and CD4⁺ T cells in both spleen and tumor-draining inguinal lymph nodes (Fig. 3A, 3B). The cps does not carry tumor-specific Ags, so it was of interest to determine whether the treatment generated T cells that recognize specific tumor Ags in the peripheral lymphoid organs. This was assessed by ELISPOT assay of CD8⁺ IFN-γ–producing T cells in response to presentation of the melanoma Ag TRP-2. There was a dramatic increase of IFN-γ–secreting CD8⁺ T cells recognizing the Kb-restricted epitope TRP-2 180–188 in both draining lymph nodes and spleens of treated versus untreated animals (Fig. 3C, 3D). These data demonstrate a systemic tumor Ag-specific CD8⁺ T cell response after cps treatment. Taken together with the results in Fig. 2, this indicates that cps monotherapy causes an early robust recruitment of innate immune leukocytes, followed by generation of activated tumor-specific lymphocytes. We and collaborators have recently shown that cps induces a strong Th1 immune response and elicits a lifelong CD8⁺ T cell immunity (7). Given that CD8⁺ T cell responses are capable of eliminating tumor masses, we next determined whether T cells were involved in tumor elimination.

To determine the relative roles of CD8⁺ and CD4⁺ T cells in tumor elimination, we evaluated cps-mediated tumor eradication and survival in conjunction with administration of depleting Abs to remove CD4⁺ or CD8⁺ T cells. Interestingly, depletion of CD8⁺ cells, but not CD4⁺ cells, abrogated the antitumor efficacy of cps (Fig. 3E, 3F), demonstrating that CD8⁺ T cells are required for the success of cps monotherapy, but CD4⁺ T cells and associated T cell help are dispensable. The potential role of NK cells was also determined by administering Abs to deplete NK1.1-expressing cells. As occurred with depletion of CD8⁺ T cells, tumor growth was similar in untreated animals and animals that were cps treated in combination with NK1.1 depletion, revealing a requirement for NK cells for successful cps monotherapy (Fig. 3G).

Tumor-specific IFN-γ–expressing CD8⁺ T cells are increased in the tumor following cps treatment (Fig. 2G). To determine the changes in IFN-γ levels after cps treatment, cytokines in tumor lysates were assayed at either 18 h or 5 d after the initial cps treatment. IFN-γ is increased 5- to 10-fold at both time points (Fig. 4A). We therefore tested the functional importance of IFN-γ by cps treatment of B16F10 tumors in IFN-γ KO mice (17). The treatment has no effect on the tumor in these mice (Fig. 4B), demonstrating a requirement for host IFN-γ for treatment efficacy.

Along with IFN-γ, IL -12 is part of the cps-stimulated immune response (7) and is strongly associated with cell-mediated cytotoxic immune responses, including antitumor responses. A requirement for IL-12 for treatment efficacy was tested using IL12p35 KO mice (18). The cps treatment did not regress established tumors in IL-12 KO mice, and the resultant tumor growth rate was not statistically different from controls over the full experimental course (Fig. 4C), demonstrating the clear requirement of IL-12 for the normal treatment response. Interestingly, the treatment did stall tumor growth during the early treatment phase when cps was most likely still present in the tumor (see days 11, 16; Fig. 4C, middle panel), a pattern not seen with IFN-γ KO mice or mice with CD8⁺ T or NK cell depletion. This shows that there are two distinct immunological phases of the treatment, an initial phase in which cps is abundant and does not require host IL-12, and a subsequent phase that no longer requires the presence of high levels of cps, but does require IL-12, reflecting an ongoing antitumor adaptive immune response.

To further dissect mechanisms by which cps treatment shrinks established tumors, we investigated chemokines by which lymphocytes could be recruited into the tumor following cps injection. Recruitment of effector T cells to tumor locations has been reported to involve interaction of the CXCR3 chemokine receptor expressed on T cells with its chemokine ligands, CXCL9, 10, or 11, generated in the tumor microenvironment (19). We assayed for expression of CXCL9 and CXCL10 in the tumors themselves 18 h or 5 d after cps treatment was initiated and found that cps treatment generated robust increases of both cytokines at each time point (Fig. 4D, 4E). To determine whether the CXCR3 axis is functionally relevant in cps-mediated tumor treatment, we assessed the efficacy of cps treatment in the presence of CXCR3 blocking Ab. CXCR3 blockade reduced, but did not eliminate the antitumor efficacy of cps treatment (p < 0.05). Although survival of CXCR3-blocked animals was significantly better than untreated controls, only 20% of the treated mice exhibited complete regression of the primary tumor when CXCR3 was blocked (Fig. 4F). Taken together, these results indicate that cps monotherapy induces elevated levels of both CXCL9 and CXCL10, and this response is rapid (apparent at 18 h) and durable (still present at 5 d). The presence of these cytokines is consistent with the partial requirement of CXCR3 signaling for efficacy.

Responses against T. gondii are reported to involve the TLR adaptor molecule MyD88 (20). The requirement for MyD88 for efficacy of treatment with cps was tested in mice lacking MyD88 (21). Surprisingly, tumors in MyD88 KO mice responded normally to the treatment (data not shown), and therefore MyD88 is not required for response to cps monotherapy. CCR5 is similarly thought to play an important role in the immune response to T. gondii by its involvement in stimulating IL-12 production (22, 23). Again, contrary to expectation, CCR5 KO mice (24) responded identically to wild-type animals to treatment of established B16F10 tumors with cps (data not shown), indicating that CCR5 signaling does not influence the therapy’s efficacy.

Across many individual experiments, 76% of wild-type mice treated with cps for dermal B16F10 melanoma developed localized or disseminated vitiligo (Fig. 5A, 5B). This indicates that the immune response that eliminated the melanoma also initiated an immune response against normal melanocytes. The vitiligo suggested that the antitumor response was systemic and may be associated with system-wide immunity and immunological memory against the tumor. To determine whether cps treatment of a primary tumor would impact a secondary tumor, mice were challenged with a secondary tumor inoculation on the day after the second injection of cps or PBS into the primary tumor. Secondary challenge inoculation was done into the skin of the flank with the primary tumor or the other flank, and treatment of the primary tumor was completed (Fig. 5C). The cps-treated mice rejected the rechallenge in every case, whereas all tumors grew on PBS-treated tumor-bearing mice (Fig. 5D, 5E) that had the primary tumor surgically removed on day 13 to better compare with standard clinical therapy and limit the size of the rapidly growing primary tumor.

To demonstrate antitumor immune memory, we rechallenged mice that had successful treatment of the primary tumor 40 or 120 d after inoculation of the primary tumor (20 or 100 d after elimination of palpable primary tumor). The majority of these mice rejected the secondary tumor challenge at day 40, and the tumors that did grow in these mice grew more slowly than in the control mice whose primary tumor was surgically removed on day 12 (Fig. 5F). Although all secondary challenge tumors grew in the 120-d mice, they grew at a significantly slower rate than surgically treated control mice (Fig. 5F).

The ability of cps, which lacks melanoma-associated Ags, to mediate antitumor immunity against B16F10 suggests that cps treatment induces tumor-specific immune responses directed against tumor Ags. Therefore, we hypothesized that cps treatment would demonstrate efficacy against other tumors. The ability of cps to treat other tumors was tested using melanomas generated in an inducible genetic melanoma system that utilizes Braf mutation and Pten loss (25), syngeneic C57BL/6 mouse-derived Lewis lung carcinoma (26), and syngeneic C57BL6 mouse-derived UpK10 ovarian cancer (11). Using a treatment approach similar to that described for B16F10, intratumoral cps administration either significantly slowed tumor growth or led to complete regression, depending on the model (Fig. 6). Thus, intratumoral cps injection is a generalizable approach that has efficacy against a variety of established tumors.

To our knowledge, this study provides the first example of the use of cps for tumor immunotherapy and the first demonstration of high-frequency successful immunological treatment of established B16F10 melanoma using either an attenuated pathogen or an immunological monotherapy. Importantly, cps accomplishes this with no infection-associated toxicity because it cannot replicate in vivo. This combination of efficacy and lack of toxicity is the essential definition of a treatment with an outstanding therapeutic index.

Approximately 20% of the world population is latently infected with T. gondii, with latent infection much higher in some areas. Latent infection with T. gondii does not impact the efficacy of intratumorally administered cps. Thus, this approach could be used to effectively treat patients, regardless of past exposure to T. gondii. This finding suggests that Abs induced by infection with wild-type T. gondii do not interfere with subsequent initial or repeated cps-based therapies. This is an advantage over viral vector-based cancer therapy strategies that often induce Abs that block infectivity following repeated administrations (27).

T. gondii has been studied previously in connection with cancer therapy. Latent infection with T. gondii slowed tumor development in mice (28). Mouse studies using acute toxoplasmosis from i.p. injection to treat cancer retarded tumor growth, but, unlike treatment with cps, it was not used to treat an established tumor, the adaptive immune system was not required, the primary tumors were not eliminated, and the approach itself was limited by toxicity of the infection (16). T. gondii extracts have been used to mature DCs in vitro to improve T cell adoptive transfer efficacy (29). This effect of T. gondii extracts is reflected in our studies using heat-killed cps, in which there was a discernable but modest slowing of tumor growth. The results reported in this study require live metabolically active cps and are not attributable to microbial constituents.

These studies show that leukocytes are preferentially, but not exclusively invaded by cps and there is a highly effective specific immune response both against the tumors and against cps. Interestingly, efficacy was achieved even though cps does not express any tumor Ags. Although it is possible that there is a Toxoplasma Ag that cross-reacts with an Ag in B16F10, the efficacy in lung and ovarian tumors argues against such a possibility. One explanation for apparent Ag spreading is that cps treatment and associated immunostimulation release adaptive immune responses that are present, but being suppressed by the tumors prior to treatment. In studies using L. monocytogenes (30–32) or S. typhimurium (4) to stimulate antitumor immunity, it was shown that expression of tumor Ags by pathogens significantly improves the antitumor immune response generated. The cps can be engineered to express Ags to increase tumor-Ag–specific responses that may further improve efficacy and immune memory development (33).

It is clear that treating dermal melanoma by immunotherapy is only superior to surgical resection if it stimulates systemic antitumor immune responses that can eliminate occult metastases. Vitiligo is generated by cps therapy and reveals an immune response against normal melanocytes. Development of vitiligo is a predictor of positive outcome in melanoma patients (34–36) and has been shown to be required for long-term T cell memory against B16F10 melanoma (37). The ability to reject or markedly impair subsequent challenges either during treatment of the primary tumor, or after the initial effector phase of the response is complete and the memory phase has begun, supports the potential for this approach to clinically impact early-stage metastatic disease. This therapeutic approach may be effective in the crucial stage in the course of the disease when it may have metastasized, but is not yet so disseminated as to be virtually incurable.

These studies demonstrate that treatment efficacy requires CD8⁺ T cells, which is not surprising, but less predictably shows that CD4⁺ T cells are not required for full efficacy. This demonstrates that the mechanism does not depend on classic T cell help. T cell help for cytotoxic T cells is associated with Th1-type Th cells and the classic Th1 cytokines IFN-γ and IL-12, which are in fact both required for efficacy using cps. This implies that other cell types are taking over the role of Th1 CD4⁺ T cells when cps is present. It is likely that one of those cell types is NK cells, which are also required for efficacy and can be a major source of IFN-γ in cancer (38).

The cps treatment efficacy required host production of IL-12, and whereas T. gondii has been previously reported to increase IL-12 levels through CCR5 signaling (22, 23), CCR5 KO mice have normal efficacy with this treatment approach. Therefore, other pathways are manipulated by cps to increase IL-12 or the IL-12 is generated through the CCR5 pathway by the genetically normal B16F10 tumor cells. Such pathways may be functional normally or may require specific changes that are associated with the presence of tumors. Similarly, MyD88 signaling has been implicated in the immune response against T. gondii, but is not needed for treatment efficacy. It appears that this therapy is working through as yet unrecognized pathways of T. gondii interaction with the immune system. It should be noted that MyD88 is broadly expressed, and because tumor cells are invaded by cps, it is possible that Myd88 in tumor cells does play a role in the treatment.

Detailed examination of tumor growth curves from the IL-12 KO mouse studies shows that treatment initially inhibits growth of the tumor while cps is present. However, the tumor again overcomes the immune response once cps is less prevalent. By contrast, other conditions that block efficacy, such as elimination of NK or CD8⁺ T cells or treatment of IFN-γ KO mice, lost all treatment impact and were identical to PBS controls. This suggests that the early response, in which cps is present in the tumor and the tumor growth halts, depends on different mechanisms than the later response during which the tumor is eliminated despite the fact that cps is no longer present. Potentially, the early response breaks immunosuppression and activates innate immune functions that support already existing, but suppressed adaptive responses, and the later response depends on expansion and recruitment of naive tumor-specific T cells.

The data show that the CXCR3 axis plays an important role in cps-mediated antitumor immune responses and is required for full treatment efficacy, but tumor growth is still significantly slowed when CXCR3 is blocked on host cells. Although this suggests that the CXCR3 axis is not the only mechanism by which effector CD8⁺ T cells are recruited to the tumor, its importance is underlined by the finding that, in humans, the presence of circulating tumor Ag-specific CXCR3⁺ CD8⁺ T cells was associated with enhanced tumor-free survival of melanoma patients (39).

The cps treatment has efficacy as a monotherapy against an orthotopic poorly immunogenic tumor. Immunological monotherapy of any sort has rarely exhibited significant efficacy, and most tumor immunotherapy approaches involve multiple, distinct immune manipulations. The demonstration of monotherapy effectiveness presented in this study is likely to be further improved by coupling with other immunological approaches, such as adoptive T cell therapy and blocking of immune checkpoints that suppress T cells. Monotherapy cps treatment of B16F10 also stands out from many other immunotherapy approaches by relying totally on manipulation of the endogenous immune response in vivo.

The efficacy against established B16F10, lack of dependence on specific Ag expression that is subject to immunoediting, response regardless of pre-exposure to T. gondii, and systemic and memory responses establish cps monotherapy of tumors as a unique potential clinical therapy. Additionally, the simplicity, safety, and ability to effectively treat multiple solid tumor types suggest that cps can be developed into a valuable clinical tool for stimulation of therapeutic antitumor immunity.

We acknowledge discussions with Jose Conejo-Garcia that contributed significantly to this project. We also acknowledge Jennifer Fields, Sandra Warner, Laurie Horne, the Dartmouth Transgenic and Genetic Construct Shared Resource, the Dartmouth Immune Monitoring Shared Resource, and the Dartmouth Flow Cytometry Shared Resource for technical support.

The authors have no financial conflicts of interest.