Targeting Stat3 in the Myeloid Compartment Drastically Improves the In vivo Antitumor Functions of Adoptively Transferred T Cells

Immunotherapies have shown promise for the prevention and treatment of various diseases (1–3). For effective vaccination against viral and other pathogen infections and for clinical benefits of cancer immunotherapies, optimizing in vivo functions of CD8⁺ T effector cells is critical (4–7). However, T-cell effector functions are often weakened in individuals with chronic infections and/or malignancies (8–10). Although adoptive T-cell therapy has shown promise for treating viral infection and cancer (11), the requirement of extensive ex vivo manipulation to expand, activate, and potentially increase homing of effector functions of T cells to tumor sites in the hosts limits its application (8, 9). Even when the T cells have been optimally engineered and activated ex vivo, their activity against tumor cells often fails to persist (12, 13). This is in part caused by the hostile tumor immunologic environment that dampens the efficacies of T cells activated ex vivo (5, 14, 15). It is therefore highly desirable to be able to efficiently upregulate effector functions of CD8⁺ T cells in vivo not only to reduce the requirement of extensive ex vivo manipulation of T cells but also to circumvent the immunosuppression associated with chronic infections and/or cancer.

We and others have recently identified signal transducer and activator of transcription 3 (Stat3) as negative regulator of Th1 immunity (16–19). In the setting of malignancy, Stat3 is persistently activated not only in tumor cells but also in tumor-associated myeloid cells as well as regulatory T cells (16, 20). Inhibiting Stat3 in either tumor cells or tumor myeloid cells can elicit Th1 antitumor innate and adaptive immune responses, which is accompanied by an increase in tumor-infiltrating CD8⁺ T cells and decrease in tumor-regulatory T cells (17). However, for potential clinical translation of these findings, it is critical to determine whether targeting Stat3 in myeloid cells can alter the effector functions of adoptively transferred CD8⁺ T cells.

It has also been shown that certain Toll-like receptor (TLR) signaling activates Stat3, which in turn constrains the magnitude of innate immune responses (21–23). Ablating Stat3 in the myeloid compartment and B cells drastically improves the efficacy of TLR9 agonist CpG-induced antitumor immune responses (24). By conjugating CpG with small interfering RNA (siRNA), we have recently developed a novel in vivo siRNA delivery technology platform achieving targeted delivery and gene silencing in myeloid cells and B cells, as well as immune activation (25). In the current study, we explore the feasibility of using CpG-Stat3siRNA to improve the effector functions of adoptively transferred CD8⁺ T cells in vivo, thereby developing an approach to alleviate the extensive ex vivo manipulations while improving the antitumor efficacies of transferred T cells.

Murine B16-F10 melanoma cells and B16 cells expressing ovalbumin (B16^OVA) were generously provided by Drs. D.M. Pardoll (Johns Hopkins, Baltimore, MD) and J. Mule (Moffitt Cancer Center, Tampa, FL), respectively. The B16 cells expressed melanoma-specific HMB-45 antigen as assessed using intracellular staining and flow cytometry (data not shown). The expression of exogenous OVA antigen and B16 cell–specific endogenous TRP2 and p15E antigens was confirmed by enzyme-linked immunosorbent spot (ELISpot) assays performed within the last 6 months. The ability of these cells to form melanoma in C57BL/6 mice and to elicit OVA-specific response was monitored.

Stat3^flox mice were kindly provided by S. Akira (Osaka University, Osaka, Japan). Ova TCR (OT-I), Rag1(ko)Momj/B6.129S7, and Mx1-Cre transgenic mice were purchased from The Jackson Laboratory. Stat3^flox and Mx1-Cre mice were crossed and treated with poly(inosinic-cytidylic acid) to obtain Stat3 conditional knockout in the hematopoietic system as described previously (26). C57BL/6 mice were purchased from the National Cancer Institute (Frederick, MD). CD11c(YFP)-Tg(BDC2.5)NOD mice were kindly provided by Dr. Chih-Pin Liu (City of Hope, Duarte, CA). Mouse care and experimental procedures were performed under pathogen-free conditions in accordance with established institutional guidance and approved protocols from the Research Animal Care Committees of the City of Hope.

B16 or B16^OVA cells (10⁶ or 2.5 × 10⁵) were injected s.c. into Rag1^−/− mice and C57BL/6 wild-type or Stat3^flox mice, respectively. CD8 or CD8^OT-I T cells (8 × 10⁶ to 10 × 10⁶) were adoptively transferred when tumors reached an average diameter of 5 mm via retro-orbital route. T cells were isolated from spleens and lymph nodes of donor mice using negative selection (EasySep, StemCell Technologies) or MACS cell separation system positive selection (Miltenyi Biotec). Fluorescent cell labeling was performed using CFSE or CMAC CellTracker (Invitrogen), according to the manufacturer's instructions.

B16^OVA tumor-bearing Stat3^flox mice received 5 μg (0.78 nmol) of phosphothioated CpG-ODN 1668 (TCCATGACGTTCCTGATGCT) injected peritumorally 5 hours before CD8^OT-I T-cell adoptive transfer. C57BL/6 wild-type mice bearing B16^OVA tumors were treated every other day with 19.2 μg (0.78 nmol) of CpG-luciferase-siRNA, CpG-scrambled-RNA, or CpG-Stat3siRNA. Adoptive T-cell transfer was performed 24 hours after the first CpG-siRNA treatment.

Cell suspensions were prepared from lymph nodes and tumor tissues as described previously, followed by staining with different combinations of fluorophore-conjugated antibodies against CD8, CD69, CD4, CD25, FoxP3, IFN-γ, and granzyme B (BD Biosciences). Flow data were acquired using FACSCalibur (BD Biosciences) and analyzed by FlowJo software (Tree Star).

Tissue sections were fixed with 2% paraformaldehyde, permeabilized in methanol, and blocked in PBS containing 10% goat serum (Sigma) and 2.5% mouse serum (Sigma). Sections were incubated overnight with primary antibodies (α-dendritic cell marker, clone 33D1, eBioscience; perforin 1, Santa Cruz Biotechnology) diluted 1:50 in PBS (supplemented with goat and mouse sera) and for 1 hour with fluorophore-conjugated secondary antibodies (Invitrogen) after washing three times with PBS. Immunofluorescent stainings were analyzed by confocal microscopy (LSM510Meta, Zeiss).

Tumor-bearing mice were anesthetized with an isoflurane/oxygen mixture. Fifteen minutes before imaging procedure, mice were given 100 μg dextran-rhodamine (Invitrogen) and 10 μg Annexin V FITC (BioVision) i.v. Extracellular matrix (ECM) emission signals were given by second harmonic generation at λ[excit] = 890 nm (Coherent Chameleon Ultra II Ti:Sa laser). For recording fluorescein and rhodamine emission, λ[excit] = 860 nm was used, and coumarin emission signals were recorded at λ[excit] = 730 nm. Labeling of CD8^OT-I cells with CMAC or CFSE CellTracker was performed according to the manufacturer's instructions. Images were acquired using an Ultima Multiphoton Microscopy System (Prairie Technologies) equipped with Prairie View software and non-descanned Hamamatsu PhotoMultiplier Tubes. Images were collected in a 512 × 512, 16-bit, TIFF format. Composite images were created using Image-Pro Plus professional imaging software (Media Cybernetics).

Cells (5 × 10⁵) isolated from tumor-draining lymph nodes (TDLN) of tumor-bearing mice as well as from lymph nodes of naïve mice were seeded into a 96-well filtration plate in the presence or absence of 10 μg/mL peptide (TRP2^SVYDFFVWL and OVA^SIINFEKL, AnaSpec; p15E^KSPWFTTL generated by the DNA/RNA and Protein Synthesis Core Facility at the City of Hope) for 24 hours at 37°C. Peptide-specific granzyme B and IFN-γ–positive spots were detected according to the manufacturer's instructions (R&D Systems, Diaclone) and manually counted using a binocular microscope.

Splenocytes of syngeneic animals were harvested and split into two populations. Target cell population was pulsed with 2 μg/mL OVA^SIINFEKL peptide for 2 hours at 37°C followed by CFSE^HI (10 μmol/L) fluorescent labeling, whereas the control cell population remained unpulsed but was labeled CFSE^LO (1 μmol/L). Equal numbers of CFSE^HI and CFSE^LO cells were mixed and adoptively transferred i.v. into tumor-bearing animals. Each animal received 20 × 10⁶ cells. CTL cytotoxic effects were analyzed by flow cytometry (FACSCalibur).

Because T-cell engagement by antigen-presenting cells is a critical step for T-cell priming (27–29), we monitored cell to cell contacts of adoptively transferred CD8^OT-I T cells and dendritic cells (DC) in B16^OVA tumor-bearing mice, which received CpG-ODN 5 hours before T-cell transfer. Intravital multiphoton microscopy (IVMPM) imaging of the TDLN was carried out 15 to 18 hours after adoptive transfer. Results from the imaging revealed significantly increased engagement of CD8^OT-I T cells by CpG⁺ myeloid cells in mice with Stat3^−/− hematopoietic cells (Fig. 1A, top, B and C). Immunofluorescent staining of frozen sections prepared from TDLNs confirmed CpG-ODN internalization by DCs (Fig. 1A, bottom). Results from flow cytometric analysis showed that adoptively transferred CD8^OT-I T cells underwent rapid activation on Stat3 ablation in the myeloid compartment, as indicated by higher CD69 expression by the T cells, relative to those from Stat3^+/+ mice (Fig. 1C). Clonal expansion of adoptively transferred CD8^OT-I T cells represents the final step of T-cell priming. We therefore tested the in vivo effect of Stat3 ablation in the myeloid compartment on transferred T-cell clonal expansion. CD8^OT-I T cells were loaded with CFSE CellTracker and transferred into B16^OVA tumor-bearing mice. T-cell expansion was assessed 16 hours after adoptive transfer. Flow cytometric analysis showed proliferation of CD8^OT-I T cells in the TDLN of mice with Stat3^−/− but not Stat3^+/+ myeloid compartment. Notably, proliferation of the transferred T cells was not detectable in the contralateral lymph node of Stat3^−/− mice, indicating that peritumoral CpG-ODN administration results in tumor-associated CD8⁺ T-cell expansion (Fig. 1D).

Both homing capacity and cytolytic activity of transferred CD8⁺ T cells are limited in tumor-bearing hosts. We therefore assessed the effect of Stat3 ablation in the myeloid compartment on CTL effector functions on CpG triggering. Immunofluorescent staining of microsections of TDLNs prepared from mice challenged with B16^OVA tumor cells showed that perforin 1 expression was considerably increased in transferred CD8^OT-I T cells in recipient mice with Stat3^−/− myeloid compartments (Fig. 2A). Notably, perforin expression was not restricted to CD8^OT-I T cells but was also observed in host lymphocytes, suggesting that perforin upregulation occurs in host CD8⁺ T cells and natural killer cells on genetic Stat3 deletion in myeloid cells and CpG administration. In addition, granzyme B and IFN-γ expression by adoptively transferred antigen-specific CD8^OT-I T cells was markedly increased on Stat3 ablation in myeloid cells (Fig. 2B). In vivo CTL killing assay further indicated that the elevated expression of granzyme B and IFN-γ led to an effector CTL population phenotype with strong and rapid cytolytic activity against tumor antigen peptide–pulsed, CFSE-labeled syngeneic splenocytes (Fig. 2C). However, the CFSE^HI target cell population that resisted CTL killing in Stat3^+/+ mice at an earlier time point (6 hours) became more susceptible at a later point (15 hours). However, unlike efficient CTL killing activity in mice with Stat3^−/− myeloid cells, CFSE^HI target cells remained resistant to CTL killing in Stat3^+/+ mice after 20 hours. Although adoptively transferred CD8^OT-I T cells are functionally active in Stat3^+/+ mice, their adaptive tolerance indicates partial T-cell anergy (30). Thus, Stat3 hinders effector CTL maturation induced by combining T-cell therapy with TLR9 agonist administration.

We next examined tumor infiltration of CMAC-labeled and adoptively transferred CD8^OT-I T cells. Results from flow cytometric analysis show improved effector CTL^CMAC+ tumor infiltration in Stat3^−/− tumor-bearing mice 24 hours after adoptive transfer (Fig. 2D), correlating with enhanced VCAM-1 expression in tumor tissue (Supplementary Fig. S1). VCAM-1 expression is thought to facilitate CD8⁺ T-cell trafficking to the tumor (31–33). Taken together, our data thus far indicate that silencing Stat3 in the myeloid compartment can improve effector CTL maturation/killing activity and tumor infiltration of transferred CD8⁺ T cells.

Because ablating Stat3 in the hematopoietic system in conjunction with CpG administration improves drastically effector functions of transferred CD8⁺ T cells, we tested CpG-Stat3siRNA as a therapeutic molecule in this setting. We monitored the cellular uptake of red fluorescently labeled CpG-Stat3siRNA in naïve transgenic mice with a yellow fluorescent DC population due to expression of yellow fluorescent protein (YFP) under control of the CD11c promoter. Fluorescent CpG-Stat3siRNA or fluorescent CpG without the siRNA moiety was injected s.c., followed by IVMPM analysis of the inguinal lymph nodes 2 hours after injection. Both CpG and CpG-Stat3siRNA were efficiently uptaken by CD11c(YFP)⁺ cells (Fig. 3A, top), indicating that the siRNA moiety does not affect cellular internalization.

Early activation of adoptively transferred CD8^OT-I T cells on treatment with either CpG-Stat3siRNA or control CpG-luciferase-siRNA was assessed in TDLNs of B16^OVA tumor-bearing mice. CpG-luciferase-siRNA–treated mice, CD8^OT-I T-cell recipient mice, and untreated mice were included as controls. Combined treatment with CpG-Stat3siRNA and CD8^OT-I T-cell adoptive transfer resulted in augmented CD8⁺ T-cell activation compared with treatment with CpG-luciferase-siRNA and T-cell transfer, CpG-luciferase-siRNA alone, or transferred CD8^OT-I T cells alone, or untreated control as shown by CD69 surface expression analysis (Fig. 3A, bottom). Moreover, Stat3 knockdown in CD11c⁺ cells achieved by repeated CpG-Stat3siRNA administration (Fig. 3B) significantly increased granzyme B and IFN-γ expression by adoptively transferred CD8^OT-I T cells (Fig. 3C). Notably, CpG-luciferase-siRNA treatment failed to induce maturation of adoptively transferred CD8^OT-I T cells into a CTL phenotype because granzyme B and IFN-γ production is not significantly changed compared with adoptively transferred CD8^OT-I T cells alone. These observations suggest that targeting Stat3 in myeloid cells by CpG-siRNA improves key effector functions of transferred CD8⁺ T cells. Finally, we examined tumor infiltration of fluorescently labeled and transferred CD8^OT-I T cells. Whereas CpG-luciferase-siRNA treatment did not improve tumor infiltration of transferred CD8^OT-I T cells, CpG-Stat3siRNA administration increased T-cell tumor infiltration (Fig. 3D).

The effects of CpG-siRNA on transferred CD8⁺ T cells in the TDLNs shown in Fig. 3 were assessed in mice during a short treatment interval. We next tested whether the improved effector functions of transferred CD8⁺ T cells induced by CpG-Stat3siRNA could lead to more potent antitumor activity in B16^OVA tumor-bearing mice with extended treatment. Again, CD8^OT-I T cells isolated from TDLNs showed significantly increased production of IFN-γ on recalling CD8^OT-I stimulation with OVA peptide, thus indicating improved effector CTL conversion on treatment with CpG-Stat3siRNA compared with CpG-luciferase-siRNA (Fig. 4A). Moreover, targeting Stat3 in the myeloid compartment using CpG-siRNA resulted in increased activation of CD8^OT-I T cells within tumors after 2 weeks of treatment, as shown by flow cytometric analysis (Fig. 4B). Notably, whereas combining CpG-luciferase-siRNA and T-cell transfer did not prevent tumor outgrowth, treatment with CpG-Stat3siRNA greatly enhanced the antitumor efficacy of adoptively transferred CD8^OT-I T cells (Supplementary Fig. S2A). Correlating with significantly elevated IFN-γ production, VCAM-1 expression increased on tumor-associated CD31⁺ endothelial cells on treatment with CpG-Stat3siRNA and adoptive T-cell transfer (Supplementary Fig. S2B).

Furthermore, the immunosuppressive CD4⁺CD25⁺FoxP3⁺ regulatory T (T_reg) cell population decreased dramatically on treatment with CpG-Stat3siRNA and CD8^OT-I T-cell transfer compared with combination of CpG-luciferase-siRNA and T-cell transfer (Fig. 4C, top), which was validated by two-photon microscopy using FoxP3-GFP mice (Fig. 4C, bottom). Although T-cell transfer alone resulted in a diminished T_reg cell population compared with untreated control, treatment with CpG-luciferase-siRNA alone did not affect the T_reg population. Combining CD8^OT-I T-cell adoptive transfer with CpG-luciferase-siRNA treatment increased the T_reg population, indicating an undesired but known effect of TLR9 agonist (34, 35). Related to this observation, it has been reported that CpG can activate Stat3, constraining antitumor immune responses (24, 36, 37). Finally, we analyzed the induction of direct tumor cell apoptosis by multiphoton microscopy in vivo. CpG-Stat3siRNA treatment for 2 weeks in mice receiving CD8^OT-I T-cell transfer resulted in markedly increased induction of tumor cell apoptosis, whereas combination of T-cell transfer and CpG-luciferase-siRNA did not affect tumor cell viability in vivo (Fig. 4D). In addition, tumor vasculature collapsed in mice treated with CpG-Stat3siRNA and CD8^OT-I T-cell adoptive transfer but was still intact in CpG-luciferase-siRNA–treated mice. Hence, targeting Stat3 with CpG-siRNA in the myeloid compartment improves the antitumor efficacy of T-cell therapy, achieving robust CD8⁺ T-cell activation and conversion into a CTL phenotype in vivo, mounting desired antitumor activity.

Because T-cell therapy is often performed in lymphodepleted setting, and to separate the effects of Stat3 inhibition in myeloid cells on host versus transferred T cells, we addressed the antitumor efficacy of CpG-Stat3siRNA administration combined with T-cell transfer in B16 melanoma tumor–bearing Rag1^−/− mice. The tumors in Rag1^−/− mice repopulated with CD8⁺ T cells underwent growth regression on CpG-Stat3siRNA treatment, whereas the tumors in the Rag1^−/− mice treated with CpG-luciferase-siRNA/CD8⁺ T cells continued to grow (Fig. 5A). In contrast, the CpG-Stat3siRNA administration alone did not completely inhibit growth of B16 tumors in wild-type mice, indicating that host CD8 T-cell population is insufficient for a desired antitumor response. Furthermore, adoptively transferred CD8 cells in B16 tumor-bearing Rag1^−/− mice treated with CpG-Stat3siRNA displayed increased expression of both granzyme B and IFN-γ on restimulation by natural B16 melanoma antigens ex vivo (Fig. 5B). Moreover, a significant B16 tumor regression was observed in Rag1^−/− mice receiving CD8 T cells and CpG-Stat3siRNA, but not in the same mice receiving control CpG-scrambled-RNA or CD8 T-cell therapy alone (Fig. 5C). The expression of both granzyme B as well as IFN-γ protein by adoptively transferred CD8 T cells isolated from TDLNs was considerably increased upon CpG-Stat3siRNA administration (Fig. 5D).

Current T-cell therapies, most notably adoptive T-cell therapies, require ex vivo activation, expansion, and/or genetic engineering to generate a desired CTL phenotype. This prolonged and extensive ex vivo requirement not only limits T-cell therapy application but also delays patient treatment. Furthermore, the tumor microenvironment poses a serious threat to dampen the effector functions of transferred T cells and constrains their persistence in the hosts (9). It is therefore highly desirable to identify approaches that can reduce or minimize the requirement for ex vivo manipulation of T cells before transfer and, even more importantly, to circumvent the immunosuppressive tumor milieu that interferes the effector functions of transferred T cells. We show here, using genetic approaches, that inhibiting Stat3 in the myeloid compartment and B cells can facilitate the achievement of this goal.

We have recently developed an siRNA delivery technology involving CpG-siRNA conjugate that facilitates siRNA uptake and gene silencing in myeloid cells and B cells. Although CpG-driven immune activation is thought to be mainly mediated by TLR9-expressing DCs, and TLR9 expression is more restricted among human DCs compared with mouse, macrophages and B cells also serve as antigen-presenting cells (38). In addition, both macrophages and B cells are important components of the tumor microenvironment, producing immunosuppressive and angiogenic/metastatic factors (39–42). At the same time, human B cells and plasmacytoid DCs, which play an important role in generating antitumor immune responses, do express TLR9 (38, 43). These findings support further development of human CpG-Stat3siRNA for the potential use in the setting of adoptive T-cell therapy.

Although our current study only tested CpG-Stat3siRNA to improve the effector functions of adoptively transferred T cells in vivo, the genetic studies presented here suggest the possible use of other Stat3 inhibitors. Currently, no direct Stat3 inhibitors are in clinical trials. This is largely due to the fact that Stat3 is a transcription factor that, unlike tyrosine kinases, lacks enzymatic activity and is difficult to drug. The use of siRNA-based therapy, therefore, is an attractive alternative approach to block Stat3 signaling. On the other hand, Stat3 is a point of convergence for many tyrosine kinase signaling pathways, and certain tyrosine kinase inhibitors in the clinic have been shown to inhibit Stat3 in the tumor microenvironment (44). In particular, sunitinib, which has been shown to reduce immunosuppressive T_reg cells and myeloid-derived suppressor cells in both patients and mouse tumor models, can inhibit Stat3 activity (45–48). In addition to Sunitinib, other tyrosine kinase inhibitors are also likely to reduce Stat3 activity, thereby enhancing antitumor immune responses. Our findings strongly suggest that Stat3 targeting, by small molecule or siRNA-based strategies, can improve the efficacy and broaden the applicability of adoptive T-cell therapy.

The authors have no conflicting financial interests.

We thank Dr. Piotr Swiderski (City of Hope, Duarte, CA) for CpG and CpGsiRNA synthesis; Dr. Chih-Pin Liu (City of Hope, Duarte, CA) for generously providing us CD11c(YFP)⁺ mice; Dr. Yong Liu for superb assistance; and the members of Flow Cytometry Core, the Light Microscopy Core, and Animal Facility at City of Hope for their contributions.

Grant Support: NIH grants R01CA122976 and R01CA146092.

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.