Patients with central nervous system (CNS) tumors are typically treated with radiotherapy, but this is not curative and results in the upregulation of phosphorylated STAT3 (p-STAT3), which drives invasion, angiogenesis, and immune suppression. Therefore, we investigated the combined effect of an inhibitor of STAT3 and whole-brain radiotherapy (WBRT) in a murine model of glioma.
C57BL/6 mice underwent intracerebral implantation of GL261 glioma cells, WBRT, and treatment with WP1066, a blood–brain barrier–penetrant inhibitor of the STAT3 pathway, or the two in combination. The role of the immune system was evaluated using tumor rechallenge strategies, immune-incompetent backgrounds, immunofluorescence, immune phenotyping of tumor-infiltrating immune cells (via flow cytometry), and NanoString gene expression analysis of 770 immune-related genes from immune cells, including those directly isolated from the tumor microenvironment.
The combination of WP1066 and WBRT resulted in long-term survivors and enhanced median survival time relative to monotherapy in the GL261 glioma model (combination vs. control P < 0.0001). Immunologic memory appeared to be induced, because mice were protected during subsequent tumor rechallenge. The therapeutic effect of the combination was completely lost in immune-incompetent animals. NanoString analysis and immunofluorescence revealed immunologic reprograming in the CNS tumor microenvironment specifically affecting dendritic cell antigen presentation and T-cell effector functions.
This study indicates that the combination of STAT3 inhibition and WBRT enhances the therapeutic effect against gliomas in the CNS by inducing dendritic cell and T-cell interactions in the CNS tumor.
Given the heterogeneous nature of gliomas, it is unlikely that a monotherapeutic strategy would induce durable responses across patients. This study combines standard-of-care radiotherapy and a blood–brain barrier–penetrant small-molecule inhibitor, WP1066, which blocks the transcriptional activity of the STAT3, which is currently in clinical trials (NCT01904123). The combination of radiation and WP1066 demonstrated a marked therapeutic response in a preclinical glioma mouse model, which was mediated by the immune system, because the therapeutic effect is lost in immune-incompetent models. By using NanoString profiling from both the CNS and the peripheral immune compartments and immunofluorescence, we found that the combination of WP1066 and radiation induces dendritic cell–T-cell interactions in the glioma microenvironment, which seems to be a requirement for a fully functional immune response. These data provide a strong rationale for clinical trial consideration in patients with glioma.
Glioblastoma (GBM) is a rapidly growing and diffusely infiltrating tumor in which patients typically survive for a median of approximately 15 months (1–3). Despite aggressive treatment including surgery, chemotherapy, and whole-brain radiotherapy (WBRT; refs. 2, 4), the propensity for tumor recurrence in patients with GBM is very high, with little improvement of patient median survival time (5, 6). Hence, it is essential to develop a more applicable adjuvant therapy combined with radiotherapy to improve the outcomes.
The STAT3 pathway is a multipotent regulator of both gliomagenesis (7) and tumor-mediated immunosuppression (8–11). STAT3 is a transcription factor, activated through phosphorylation induced by a variety of signals in a variety physiologic processes. In the tumor microenvironment, the EGFR and the IL6 signaling pathway play important roles in activating STAT3 (12, 13). Aberrant activated STAT3 inhibits the production of proinflammatory cytokines that support the maturation of dendritic cells and thereby influences the generation of antigen-dependent T cells (14, 15). STAT3 has been shown to be abnormally activated in diverse immune cells within the glioma microevironment such as T cells, NK cells, neutrophils, and different myeloid cell populations, resulting in profound immune suppression. Thus, in the tumor microenvironment, overactive STAT3 creates an immunologic niche supporting tumor cells and preventing immunosurveillance.
WBRT can induce a proneural to mesenchymal transition, with associated invasiveness and resistance to temozolomide. These changes are associated with the activation of the STAT3 pathway, and this transition could be blocked with upstream STAT3 inhibitors. As such, STAT3 blockade has been proposed to prevent the emergence of therapy-resistant mesenchymal GBM tumors at relapse (16). Various studies have shown different effects of radiation on the STAT3 pathway, depending on the dosing and timing. In a more recent study, low-dose radiotherapy has been proposed as a way to actually inhibit the STAT3 signaling pathway (17), whereas others have shown that irradiation with higher doses promotes the phosphorylation of STAT3 in a dose- and time-dependent manner (18). Radiotherapy has also been reported to lead to increased expression of STAT3 in a variety of solid tumors including melanoma (19, 20), lung (21), breast (12), and head and neck tumors (22). However, radiation can have positive antitumor immunostimulatory effects. For example, damage-associated molecular patterns such as those of HBMG1 and adenosine triphosphate being released from dying tumor cells can trigger the maturation of dendritic cells plus antigen uptake, thereby resulting in T-cell activation and recruitment to kill specific tumor cells (23, 24). Radiotherapy is also known to induce antigen shedding (25).
WP1066 is a blood–brain barrier–penetrant caffeic acid analogue that blocks the nuclear translocation of p-STAT3 (26) and is now being used in a clinical trial (NCT01904123). WP1066 has demonstrated potent direct cancer cell cytotoxicity against a wide variety cancers and therapeutic in vivo efficacy against gliomas (7, 8, 27, 28), leukemia (29), melanoma (30–32), squamous cell cancer (33, 34), renal cell cancer (35), non–small cell lung cancer (36), and breast cancer (37). Notably, many of these malignancies are treated with radiotherapy as a standard of care. WP1066 also has significant immunomodulatory properties, including on innate immune cells. Specifically, WP1066 can induce the expression of costimulatory molecules on peripheral macrophages and tumor-infiltrating microglia ex vivo from patients with glioblastoma, cell types that are typically refractory to modulation with other types of immune therapeutics. WP1066 treatment of the peripheral blood from glioblastoma patients who are immunologically anergic triggers marked production of proinflammatory cytokines, induces T-cell proliferation and effector responses, and inhibits regulatory T cells (Treg; ref. 8). Furthermore, the immunosuppressive properties of glioblastoma cells are significantly diminished upon treatment with either siRNA targeting STAT3 or with physiologic doses of WP1066 (9, 11). Collectively, these data indicate that WP1066 can reverse both innate and adaptive tumor-mediated immunosuppression and that it has direct antitumor effects. Hence, we hypothesized that abnormal activation of STAT3 would be a potential therapeutic target in the radiation resistance and that STAT3 blockade would be a potent inducer of antitumor immune cytotoxicity. To evaluate therapeutic synergistic activity and host immune response, an immunocompetent glioma model system was used (38, 39). The combined treatment of WP1066 and WBRT demonstrated increased median survival time, induction of immunologic memory, and increased antigen presentation and T-cell activation during the therapeutic window within the CNS glioma microenvironment.
Materials and Methods
In vivo murine tumor models
All animal experiments were conducted in compliance with the guidelines for animal care and use established by The University of Texas MD Anderson Cancer Center (Houston, TX) under the IACUC-approved protocol (00001176-RN00). The murine glioma GL261 cell line was purchased from the NIH (Bethesda, MD). These cells were maintained in DMEM (Life Technologies), supplemented with 10% FBS, 1% penicillin/streptomycin, and 1% l-glutamine, at 37°C in a humidified atmosphere of 5% CO2 and 95% air. The cells were cultured and numerically expanded for up to 2 weeks before intracranial implantation and tested in the week before the injection for Mycoplasma contamination (MycoAlert, Lonza).
To induce intracranial tumors in C57BL/6J or nude mice, GL261 cells were collected in logarithmic growth phase, loaded into a 25-μL syringe (Hamilton) and injected 2 mm to the right of bregma and 4 mm below the surface of the skull at the coronal suture using a stereotactic frame (Stoelting). The intracranial tumorigenic dose for GL261 cells was 5 × 104 in a total volume of 2 μL. Mice were randomly assigned to control or treatment groups (n = 6–10/group) after tumor implantation for the intracranial model systems. The animals were observed daily, and when they showed signs of neurologic deficit (lethargy, failure to ambulate, lack of feeding, or loss of >20% body weight), they were compassionately killed. These symptoms typically occurred within 48 hours before death. Their brains were removed and placed in 4% paraformaldehyde and embedded in paraffin.
WP1066 (26), which blocks p-STAT3 (8), was synthesized and supplied by Dr. Waldemar Priebe (of MD Anderson). WP1066 does not influence JAK2 kinase activity at concentrations up to 10 μmol/L based on KINOME scan profiling (Supplementary Table S1), and this compound has a selectivity score S (35) of 0.037. The IC50 of WP1066 for GL261 is 4.91 μmol/L (Supplementary Fig. S1A), whereas immunomodulation occurs at 1 μmol/L (8, 31). For in vivo treatment, the mice were treated via oral gavage with WP1066 (60 mg/kg) in a vehicle of DMSO/PEG300 (20 parts/80 parts) or vehicle control on a Monday/Wednesday/Friday schedule for 3 weeks, starting on day 7 after glioma implantation when gliomas have been shown to be established in the brain (40) and consistent with the treatment window used for evaluating other immunotherapeutics in this model (41, 42), which results in a serum circulating level in the range of approximately 1 to 3 μmol/L (Supplementary Fig. S1B). For WBRT, mice were anesthetized using isofluorane, and the whole brain was irradiated at a 2 Gy dose with an opposing lateral plan using a 15-mm collimator. The dosing and schedule of the radiation were optimized so as to not be curative, thus enabling an assessment of whether an additive or synergistic effect could be detected (Supplementary Fig. S2).
IC50 cell proliferation/survival assay
GL261 cells were seeded in triplicate at a density of 2,000 cells per well in 96-well culture plates and were treated with WP1066 at increasing concentrations of 0, 1.56, 3.13, 6.25, 12.5, and 25 μmol/L. After 72 hours of treatment, 25 mL of 5 mg/mL dimethyl thiazolyl diphenyl tetrazolium salt (MTT, Sigma-Aldrich) solution was added to each well, and the cells were cultured for 3 hours at 37°C in a humidified atmosphere of 5% CO2 and 95% air. The cells were lysed with 100 μL/well of lysing buffer (50% dimethylformamide, 20% SDS, pH 5.6) and incubated at room temperature overnight. Cell proliferation and viability were evaluated by reading the OD at 570 nm, and the IC50 was calculated.
Mice were imaged at the MD Anderson Small Animal Imaging Facility using a 7-Tesla (T) 30-cm horizontal bore magnet (Bruker Biospin MRI). Each mouse was anesthetized with 2% isoflurane during imaging. For tumor detection, T2-weighted images (21 transverse slices with a thickness of 0.75 mm and taken in a field of view of 30 × 22.5, with a matrix size of 256 × 192 pixels, for a resulting in-plane resolution of 0.117 μm) were acquired using a RARE (rapid acquisition with relaxation enhancement) sequence, with a repetition time (TR) of 3000 ms and an echo time (TE) of 57 ms. The tumor volume was determined by using the software ImageJ.
HPLC detection and quantification of serum concentration of WP1066
Mice were treated via oral gavage with WP1066 (60 mg/kg) on Monday, Wednesday, and Friday, and blood was collected via terminal cardiac puncture at various time points into K2EDTA blood vacutainers (BD Biosciences): day 1: 0 (pretreatment), 0.5, 1, 2, 4, 8, and 12 hours; day 2: 0, 0.5, 1, 2, and 12 hours; day 8: 0 (pretreatment), 0.5, 1, 2, 4, 8, and 12 hours; and day 9: 0, 0.5, 1, 2, and 12 hours. The blood was centrifuged (1,000 × g, 15 minutes, 4°C) and the plasma immediately transferred and frozen (−70°C or below). A validated method for detecting the concentration of WP1066 was conducted by IIT Research Institute (Chicago, IL).
KINOMEscan profiling of WP1066 – scanMAX
KINOMEscan profiling of WP1066 at concentrations of 1 and 10 μmol/L were assessed using Assay scanMAX performed by DiscoverX (https://www.discoverx.com/services/drug-discovery-development-services/kinase-profiling/kinomescan).
In vivo experiment for NanoString gene expression analysis
GL261 cells were injected intracranially into C57BL/6 mice and treated as described above. MRIs were taken at various time points to verify the presence of tumor. On day 15, the mice were euthanized, and their spleens were removed and their brains collected after cardiac perfusion with PBS. Immune cells were isolated from the brains by Percoll gradient density centrifugation, followed by an “untouched” T-cell selection (Mouse Pan T Cell Isolation Kit II, MACS Miltenyi Biotec). Afterwards, RNA was isolated from both the T-cell fraction and the flow-through non–T-cell fraction (CD11b+, CD11c+, CD19+, CD45R+, CD49b+, CD105+, MHC class II+, and Ter-119+), which would include the antigen-presenting cells (RNeasy Plus Mini Kit, Qiagen) for NanoString gene expression analysis. For the characterization of the T-cell fraction and the flow through, the non–T-cell fraction (other immune cells) was first stained with fixable viability dye eFluor 780 to exclude dead cells (Thermo Fisher Scientific), followed by staining with anti-mouse CD45 BV510 (clone 30-F11), anti-mouse CD3 PerCP.Cy5.5 (clone 17A2), anti-mouse CD11b PE (clone M1/70), anti-mouse CD11c APC (clone N418), anti-mouse CD19 BV421 (clone 6D5), anti-mouse CD49b PE/Cy7 (clone DX5; all BioLegend). Afterwards, cells were fixed with fixation buffer (BD Biosciences) and measured using FACS Celesta (BD Biosciences). The data analysis was done with FlowJo software.
Ex vivo flow analysis of tumor-infiltrating immune cells
GL261 cells were injected intracranially into C57BL/6 mice and treated as described above. On day 17, the mice were euthanized, and their brains were collected after cardiac perfusion with PBS. Immune cells were isolated from the brains by Percoll gradient density centrifugation. Cells were then first stained with fixable viability dye eFluor 780 to exclude dead cells (Thermo Fisher Scientific). For the distinction of dendritic cells and microglia, cells were stained with anti-mouse CD45 BV421 (clone 30-F11), anti-mouse CD11c APC (clone N418), anti-mouse CD103 BV605 (clone 2E7), anti-mouse CD11b PE (clone M1/70), anti-mouse MHC-II BV785 (clone M5/114.15.2; all BioLegend), and anti-mouse Tmem119 (28-3; Abcam), followed by staining with the secondary goat anti-rabbit AlexFluor488 antibody. Afterwards, cells were fixed with fixation buffer (BD Biosciences) and measured using FACS Fortessa (BD Biosciences). The data analysis was done with FlowJo software.
RNA (200 ng) at a concentration of 40 ng/μL in a total volume of 5 μL was prepared for NanoString assay analysis with the immune-specific Gene Array Kit (NanoString Technologies, Inc.). Sample preparation and hybridization were performed for the assay according to the manufacturer's instructions. Briefly, RNA samples were prepared by ligating a specific DNA tag (mRNA-tag) onto the 3′ end of each mature mRNA, and excess tags were removed via restriction enzyme digestion at 37°C. After processing using the mRNA Sample Preparation Kit, the entire 10-μL reaction volume containing mRNA and tagged mRNAs was hybridized with a 10-μL reporter CodeSet, 10 μL of hybridization buffer, and a 5-μL capture ProbeSet (for a total reaction volume of 35 μL) at 65°C for 16 to 20 hours. Excess probes were removed using two-step magnetic bead–based purification with an nCounter Prep Station. The specific target molecules were quantified using an nCounter Digital Analyzer by counting the individual fluorescent bar codes and assessing target molecules. The data were collected using the nCounter Digital Analyzer after obtaining images of the immobilized fluorescent reporters in the sample cartridge using a charge-coupled device camera. These data were then normalized to mRNA gene expression data for all 770 immune-related genes using the NanoStringNorm R package (version 1.1.17). The cluster analyses were used to determine deregulated genes between different treatment and control groups by multigroup comparison using Qlucore software. The genes specific to certain immune cell types and functional signaling pathways were categorized on the basis of the attached kit manual.
Gene set enrichment analysis
Data from the NanoString experiment were loaded into an nCounter (NanoString Technologies) to generate corrected counts using internal standards. For each of the categories provided from the NanoString manual, a gene set was generated. Corrected counts were then loaded using software Gene set enrichment analysis (GSEA) 4.0.3 for all four samples, and the GSEA analyses were run on all gene sets. These were used as our gene set database with 1,000 permutations and with no gene collapsing or remapping. Heatmaps, enrichment scores, and correlations were all produced in GSEA (43). The analysis failed for some of the gene sets when the gene set was either too small or did not have enough genes expressed in the dataset.
A tyramide signal amplification protocol was used to show the CD11c expression in the brain tumors. The CD11c antibody (Abcam ab219799) was validated using IHC. PerkinElmer DAPI diluted to 1:75 was used as a nuclear counterstain. Slides were incubated for 2 hours at 60°C, dewaxed in xylene (three times × 10 minutes), and rehydrated through a graded series of ethanol solutions 100% (two times × 5 minutes), 90% (two times × five minutes), 80% (1 time × 5 minutes), 70% (one time × 5 minutes) followed by a PBS rinse (two times × 5 minutes). The slides were treated with hydrogen peroxide and methanol (3% w/v) for 20 minutes at room temperature then rinsed again with PBS (two times × 5 minutes). Antigen retrieval was done using the BioGenex EZ-Retriever System V.3 microwave for one cycle of 15 minutes at 95°C with a pH 9 buffer. Slides were left to cool for 30 minutes and then rinsed again in PBS (2 × 5 minutes). PerkinElmer ready-to-use blocking solution was applied for 35 minutes, and then the slides were incubated with the primary antibody diluted in the same blocking solution (1:100) overnight. The second day, the slides were washed with PBS mixed with 0.1% Tween 20 (TPBS; three times × 5 minutes). Then, slides were incubated in the PerkinElmer HRP polymer solution for 30 minutes at room temperature (∼2–3 drops/slide), then washed again with TPBS (3 × 5 minutes). Slides were incubated in the fluorochrome solution (Opal reagent 570 diluted with the amplification solution to 1:100) for 5 minutes at room temperature, and then washed with TPBS (3 × 5 minutes). PerkinElmer DAPI diluted in PBS to 1:75 was used as a nuclear counterstain for 15 minutes at room temperature, and then, slides were washed with PBS (1 × 5 minutes) and mounted with Dako fluorescence mounting medium and 22 × 50 No 1.5 thickness coverslip glass. Slides were air-dried, labeled, and stored at 4°C.
For dual CD3 and CD11c immunofluorescence analysis, slides were baked in the oven at 65°C for 1 hour, dewaxed with xylene (three times × 10 minutes), and rehydrated through a graded series of ethanol solutions (100% 1 × 10 minutes; 95% 1 × 10 minutes; and rinse in 70%). After rehydration, slides were rinsed in distilled water and were treated with 1% H2O2 in 10% methanol for 15 minutes. Slides were then rinsed in distilled water and then in the appropriate pH 9 antigen retrieval buffer for the CD11c marker. Slides were treated in the EZ Retriever microwave for 15 minutes at 95°C, and then were left to cool down at room temperature (15–30 minutes). Slides were rinsed in distilled water followed by TPBS (1%). Blocking was done using Dako ready-to-use reagent. The primary antibody CD11c (Abcam, ab219799, 1:75) was added on to the slides and incubated overnight at 4°C. Slides were rinsed in TPBS three times × 2 minutes at room temperature. Slides were incubated in PerkinElmer Polymer horseradish peroxidase for mouse and rabbit (HRP Ms + Rb) for 20 minutes at room temperature then rinsed again with TPBS three times × 2 minutes followed by incubation for 6 minutes with Opal Fluorophore Working Solution on each slide (fluorophore 570, 1:100), and then rinsed with TPBS three times × 2 minutes. The process was repeated again for the addition of the second antibody (CD3; Abcam, ab16669, 1:600). Slides were treated again with the EZ retriever microwave for one cycle of 15 minutes at 95°C with Agilent pH 6 Ag retriever buffer. Microwave stripping was followed by the same steps as above, with the washing in between: blocking, addition of the CD3 antibody (Abcam, ab16669, 1:600, overnight at 4°C), HRP treatment, and Opal 690 fluorophore diluted 1:100. Finally, slides were counterstained with DAPI (10 μL in 5 mL PBS) for 15 minutes, and then they were mounted with DAKO fluorescence mounting medium.
The dual staining was quantified manually. Slides were scanned with the Vectra Polaris, and fields of interest containing the tumor were stamped in Phenochart. The positives were sorted into two categories: dyads of one CD3+ cell interacting with one CD11c+ cell within a distance of 15 nm and clusters defined by two or more CD11c+ cells and two or more CD3+ cells. Data were merged in ExCel, analyzed for significance level, and plotted. High-resolution images presented in Fig. 5A–D were captured using an Andor Revolution WD Spinning disk confocal.
IHC for p-STAT3
Formalin-fixed, paraffin-embedded brain tumor slides were incubated for 1 hour at 60°C, dewaxed in xylene, and rehydrated in a graded ethanol series (100%, 95%, 70%) followed by water. Antigen retrieval was performed using a citrate buffer at pH9 in a pressure cooker at 120°C for 12 minutes, followed by two washes in 1× TPBS buffer mixed with 0.1% Tween 20 for 5 minutes. Peroxidase activity was blocked with 10% methanol and 1% hydrogen peroxide for 30 minutes followed by a wash. The slides were then blocked with protein blocker (Dako) for 15 minutes. The primary antibody was added to the slides and incubated overnight at 4°C (p-STAT3, Abcam: ab76315). Three washes were done followed by incubation with the secondary antibody for 30 minutes at room temperature. The DAKO DAB Kit was used for color development (10–60 seconds depending on the antibody), then counterstained with hematoxylin (25 seconds) and bluing buffer, rehydrated, and coverslipped.
Kaplan–Meier product-limit survival probability estimates of overall survival were calculated (44), and log-rank tests (45) were performed to compare overall survival between treatment groups and the control arm. To compare the amount of dyads and clusters between the different treatment groups, the mean number of dyads and cluster per field were computed. Two-sample t tests were performed on the mean number of dyads and mean number of clusters among treatment groups.
STAT3 pathway blockade in combination with WBRT increases the incidence of long-term survivors in mice with intracerebral gliomas
C57BL/6J mice bearing intracranial GL261 tumors were treated with WBRT and/or WP1066. The mice were randomized to receive (i) WP1066 (60 mg/kg for 3 weeks on M, W, F); (ii) radiation at 2 Gy; (iii) radiation + WP1066; and (iv) no treatment/control (Fig. 1A). Subtherapeutic cancer cell doses of WP1066 were used to look for synergy with irradiation. WP1066 treatment was started on day 7 after tumor implantation, and WBRT was administered on day 10 (Fig. 1A). In two different experiments (Supplementary Fig. S3), long-term durable survival was observed only with the combination of radiation and WP1066 (Fig. 1B). More specifically, the control mice had a median survival time of 23 days, WP1066-treated mice had a median survival time of 27 days (P = 0.1178 vs. control), WBRT-treated mice had a median survival time of 28 days (P = 0.0320 vs. control) and WBRT + WP1066-treated mice had a median survival time of 32 days. Forty percent were long-term survivors (>60 days after tumor implantation), which was statistically significant relative to controls (P < 0.0001), WP1066 monotherapy (P = 0.0004), and WBRT alone (P = 0.0035). MRI on day 17 after tumor implantation confirmed that tumors were present in all groups. Four of seven imaged mice from the combined treatment group showed very low tumor burden compared with the other groups, indicating that the combination therapy was suppressing tumor growth. (Fig. 1C). In the combined treatment group, the long-term survivors did not show evidence of persistent tumor on MRI (Fig. 1D). Rechallenge of the tumors in the contralateral hemisphere demonstrated protective immunity (Fig. 1E). IHC analysis of p-STAT3 during the therapeutic window (Supplementary Fig. S4) showed reduced staining intensity for p-STAT3, especially in the WBRT + WP1066-treated mice.
Therapeutic activity of STAT3 pathway blockade in combination with WBRT requires the immune system
To ascertain whether the therapeutic effect of the combination required an intact immune system, the prior therapeutic experiment was repeated using nude (athymic) mice (Fig. 1F). An equivalent number of treatment/cycles were administered to the nude mice relative to the immunocompetent model, and treatment failed to demonstrate a therapeutic response (Fig. 1G), suggesting that the immune system has a mechanistic role in the group receiving the combined therapy. As such, additional analysis was conducted to verify the role of the immune system.
WP1066 combined with WBRT modulates immune responses directly in the CNS glioma microenvironment
Because we had evidence that the immune system was required for response to the combination therapy, we next evaluated alterations in immune responses in both the peripheral and glioma microenvironment. Day 15 was selected for the therapeutic window analysis (Fig. 2A) after we documented that the tumors were large enough to analyze (Fig. 1C) but before the animals were moribund and treatment was failing (Fig. 1B). The T cells (CD3+) and other immune cells (CD11b+, CD11c+, CD19+, CD45R+, CD49b, CD105+, MHCII+, and Ter-119) were isolated from the brains and spleens, respectively, that were pooled from three to four mice from groups that were either untreated, treated with WP1066, WBRT, or WP1066 + WBRT. T-cell purity was approximately 73%, and the other immune cells consisted mostly of CD11b+ monocyte/macrophages and CD11c+ dendritic cells (Fig. 2B). To screen for mechanisms that could contribute to the observed survival benefit mediated by the combinatorial treatment, gene expression analysis was performed using NanoString, which profiled a total of 770 genes (Supplementary Fig. S5). When we compared the immunologic responses of the various treatment groups, we were surprised to find that the most robust immunological responses, such as IFN-induced responses, IL1-associated genes, and the toll-like receptor pathway were most robustly upregulated in the brain's immune cells rather than in the peripheral spleen cells (Fig. 2C). Peripheral immune monitoring has not been correlated with therapeutic responses before, and these data suggest that the brain compartment may be a more appropriate location for monitoring antitumor immune responses. As such, we focused our analysis on the immune cell responses and interactions within the brain.
WP1066 combined with WBRT globally reprograms the immune responses in the tumor microenvironment
Immune gene sets were annotated and categorized based on their known functions. For each of these categories, we performed GSEA to compare the combinatorial treatment group with the control and monotherapy groups. For both datasets, [T cells (Fig. 3A) and other immune cells (Fig. 3B)], the normalized enrichment score for each gene set was ranked from high to low to clarify the mechanisms that could have a significant influence in the response to the treatment combination. For example, genes associated with the generalized functions of immune cells (such as autophagy, cancer progression, the cell cycle, senescence) and genes associated with specific immune cell populations such as Tregs, B cells, Th2 cells, and NK cells were not enriched in the combination cohort, which indicated that there is specificity in the immune functions that are upregulated in the glioma microenvironment (Fig. 3, Supplementary Fig. S6). Whereas within the significantly enriched gene set, many were associated with T-cell–dendritic cell interactions such as antigen processing, MHC-I/-II, dendritic cell function, phagocytosis, and T-cell proliferation. Indeed, the only gene set significantly enriched in the other immune cell population was dendritic cell function (Fig. 3B and C). These data suggest that antigen presentation is a requirement within the tumor microenvironment of the CNS for full antitumor immune-mediated activities mediated by the combination of WP1066 and WBRT.
WP1066 combined with WBRT triggers dendritic cell–T-cell immunologic interactions in the glioma microenvironment
Dendritic cells usually reside in lymphoid organs, and their presence in gliomas has only recently been noted (46). The NanoString analysis strongly implicated this immune cell population in the therapeutic activity of the combination in the glioma microenvironment. To validate this key NanoString finding, we performed IHC analysis of the glioma during the therapeutic window. Glioma-bearing mice (n = 16) were either untreated (n = 4) or treated with WP1066 (n = 4), WBRT (n = 4), or the two in combination (n = 4). Immunofluorescence detection for CD11c+ cells demonstrated that these cells are abundant at the invasive edge (Fig. 4A) and are diffusely present throughout the glioma (Fig. 4B). Two of four mice in the WP1066 plus WBRT combination group achieved high CD11c positivity (defined as greater than 20% mean positivity), whereas the rates for the other groups were: control group, 0/4; WBRT, 1/4; and WP1066, 0/4 (Fig. 4B and C), further validating the results in the NanoString dataset indicating that the dendritic cells may be markedly enriched in the glioma microenvironment in mice treated with WP1066 + WBRT.
To ascertain whether the T cells and CD11c+ cells are directly interacting in the glioma microenvironment, as would be implied by the NanoString analysis, we performed dual fluorescence IHC analysis for both immune populations (Supplementary Fig. S7). A positive interaction was scored when these immune cells were within a 15-nm distance of each, indicating that the antigen-presenting dendritic cell is triggering T-cell activation (47, 48). Antigen-specific immune responses are also known to occur in clusters of dendritic cells and T cells, which are required for T-cell proliferation and effector responses (49). Tight CD11c+–T-cell intercellular interactions between individual cells (dyads; Supplementary Video S1; Fig. 5A–C) were much more frequently observed in all three treatment groups than in the control (Fig. 5E and F), indicating immunologically synapses were occurring. In addition to dyad formation, clustering interactions (Supplementary Video S2; Fig. 5D) were observed almost exclusively in the combinatorial treatment group (Fig. 5E and G; control vs. WP1066 P = 0.5696; control vs. WBRT P = 0.6951; control vs. WBRT + WP1066 P = 0.0188; WP1066 vs. WBRT P = 0.3256; WP1066 vs. WBRT + WP1066 P = 0.0140; WBRT vs. WBRT + WP1066 P = 0.0261).
To further show that the CD11c cells were dendritic cells and not microglia, immune cells were isolated from all treatment groups during the therapeutic window and analyzed via multicolor flow cytometry. Dendritic cells (defined as CD45+, CD11c+, MHC-II+, and CD103+ or CD11b+, Tmem119 negative) were almost exclusively in the CD45 high–positive population (Fig. 6A), whereas microglia (defined as CD45+ CD11b+ Tmem119+) were only detected in the CD45 intermediate population (Fig. 6B). Direct comparison between immune cells isolated from naïve brains (Fig. 6B) and the WP1066 + WBRT treated GL261 tumor-bearing brains (Fig. 6A) showed a marked increase of dendritic cells. Back gating of CD11c+ cells further confirmed that CD11c+ cells are not microglia cells (Fig. 6C). Quantification of CD103+ CD11c+ CD45 high dendritic cells, which have been described as a rare population but one remarkably capable of stimulating cytotoxic T-cell responses (50) in the tumor microenvironment, demonstrated a significant increase in the different treatment groups compared with the control, with the strongest increase being in the combination group (Fig. 6D). Cumulatively, these data indicate that dendritic cell–T-cell interactions in the glioma microenvironment are contributing to the therapeutic effect of this particular combination.
Our NanoString and immunofluorescence in vivo data indicate that the combination of WBRT and STAT3 inhibition triggers dendritic cell and T-cell interactions in the glioma microenvironment, including antigen presentation and T-cell activation, a first report to the best of our knowledge. Whereas both monotherapies increased singular dendritic cell–T-cell interactions in the tumor microenvironment by an amount similar to what was seen with the combinatorial treatment, extensive clustering of dendritic cells and T cells (which is an early event during immune activation) covering large parts of the whole tumor area, was uniquely observed only in the tumor microenvironment of the combinatorial treatment group, which is probably the key factor in the therapeutic effect. During the initial antigen presentation events, immune checkpoints are not yet significantly upregulated (51), and thus, T cells are free to exert their effector responses, including eradication of the tumor. We postulate that if the antitumor immune activation events only occur in the peripheral lymph node, especially repeatedly and chronically, with subsequent trafficking of the effector cell to a profoundly immunosuppressive CNS glioma microenvironment, that the T cell is rendered exhausted and unable to recover effector functions (52). This immunosuppression is especially problematic when STAT3 is activated, because it is the key transcriptional factor that renders tumor-infiltrating dendritic cells dysfunctional (53). Because WBRT triggers the activation and influx of the dendritic cells (54–56), and because STAT3 blockade reverses the immunosuppressive glioma microenvironment (8, 11, 28), de novo antigen presentation, and T-cell activation can occur unencumbered in this scenario. Alternatively, the observed dendritic cell and T-cell interactions may represent and support the notion of the second-touch hypothesis. In 2014, Klaus Ley put forth the second-touch hypothesis stating that full T-cell activation requires a second interaction with an antigen-presenting cell in the nonlymphoid, antigen-expressing target tissue (57). This initially marginalized concept seems to be supported by the findings in this article.
It is unlikely, given the marked heterogeneity of tumors, that single monotherapeutic strategies will provide uniform, consistent therapeutic responses among all patients. A rational strategy is to build upon standard-of-care treatment strategies such as radiation therapy. Our preclinical data support clinical trial consideration of the combination of radiation with a STAT3 inhibitor in glioma patients. But even more importantly, our study demonstrated a unique mechanism of activity, global reprogramming of the immune system within the tumor microenvironment. Immune clearance of a tumor is not mediated by a single immune cell population, chemokine, cytokine, or pathway. Rather the combinatorial treatment impacts different immune cell populations including the CD11c+ dendritic cell, which has not been fully appreciated previously in the glioma microenvironment. It is through global immunologic reprogramming, that we are able to observe the final end result of increased dendritic cell infiltration, maturation, and T-cell interaction. Future studies will be directed toward further analysis of the immunologic synapse, including further defining of the participating T-cell and dendritic cell subsets. Although we observed significant dendritic cell–T-cell interactions in the combinatorial treatment group, we did not resolve whether these interactions are tumor-antigen specific. The lack of known tumor antigens in mouse gliomas and the availability of specific tetramers makes this a challenge for the entire field. Although there are several antigen models available, they are highly immunogenic, which is not reflective of the immunobiology of gliomas.
Our immune-monitoring data in the preclinical models strongly suggest not only that peripheral immune monitoring would be of limited utility in the context of human clinical trials but also that this needs to be done with consideration of the actual tumor microenvironment. An expansion phase 0/window-of-opportunity cohort should be considered to verify that radiotherapy plus WP1066 also induces dendritic cell trafficking, maturation, antigen presentation, and T-cell activation/cytotoxicity during the therapeutic window in the context of treated human subjects. This is a plausible strategy in that we have performed similar studies in patients with glioblastoma receiving immune checkpoint inhibitors prior to surgical resection. A number of interesting therapeutic targets were also found to be upregulated in this study after administration of the combination of WP1066 and WBRT, including CD274 (PD-L1), which could also be considered in the future for clinical trial implementation.
Disclosure of Potential Conflicts of Interest
Y. Hashimoto reports grants from NCI during the conduct of the study. W. Priebe reports grants, personal fees, and other from Moleculin Biotech (shareholder and chair of scientific advisory board) during the conduct of the study; grants and other from CNS Pharmaceuticals (shareholder and chair of scientific advisory board), other from WPD Pharmaceuticals (shareholder and chair of scientific advisory board) and Houston Pharmaceuticals (shareholder) outside the submitted work; and multiple patents regarding the creation of WP1066, including its combinatorial use with other strategies, and methods for detection in serum and brain, that are held by MD Anderson Cancer Center and licensed to Moleculin. R.J. Zielinski reports personal fees from Moleculin Biotech Inc. (consultant agreement) during the conduct of the study, as well as other from CNS Pharmaceuticals (Shareholder) outside the submitted work. J.P. Long reports grants from Brockman Foundation during the conduct of the study. J. DeGroot reports grants or research support from Sanofi-Aventis, AstraZeneca, EMD-Serono, Eli Lilly, Novartis, Deciphera Pharmaceuticals, and Mundipharma, payment from Celldex, Deciphera Pharmaceuticals, AbbVie, FivePrime Therapeutics, GW Pharma, Carthera, Eli Lilly, Kadmon, Boston Biomedical, Taiho Pharmaceuticals, Kairos Venture Investments, Syneos Health, Monteris, Agios, Mundipharma Research, GenomiCare, Blue Earth Diagnostics, Del Mar, Pharmaceuticals, Insightex, Voyager Therapeutics, Merck & Co., Tocagen, Bioasis Technologies, ResTORbio (paid consultant), Genentech, Celldex, Foundation Medicine, Novogen, Deciphera, AstraZeneca, Insys Therapeutics, Merck, Eli Lilly, Novella Clinical, Karyopharm Therapeutics, Blue Earth Diagnostics, Kiyatec, Vanquish Oncology, Orsenix, Insightex, Prelude Therapeutics, Debiopharm Therapeutics, and Janssen Global Services (advisory board), other relevant financial or material interests from VBL Therapeutics, Novella, VBI Vaccines, stock ownership of Ziopharm Oncology, and employment of spouse at Gilead (Ziopharm Oncology). E.P. Sulman reports group, personal fees, and nonfinancial support from Novocure, personal fees from Merck, Blue Earth Diagnostics, BrainLab, Zai Labs, and Physicians' Education Resource, and grants and personal fees from AbbVie outside the submitted work. A.B. Heimberger reports grants from NIH, CPRIT, Ben and Catherine Ivy Foundation, Brockman Foundation, and MD Anderson Moonshot during the conduct of the study, as well as a patent regarding combinatorial use of WP1066 with interferon, but is not a party to the patent's subsequent licensing to Moleculin. No potential conflicts of interest were disclosed by the other authors.
M. Ott: Conceptualization, investigation, methodology, writing-original draft, project administration. C. Kassab: Formal analysis, investigation, methodology. A. Marisetty: Conceptualization, investigation, writing-original draft. Y. Hashimoto: Conceptualization, investigation, methodology, writing-original draft. J. Wei: Conceptualization, formal analysis. D. Zamler: Formal analysis. J.-S. Leu: Formal analysis. K.-H. Tomaszowski: Investigation. A. Sabbagh: investigation. D. Fang: Visualization. P. Gupta: Conceptualization. W. Priebe: Conceptualization, resources. R.J. Zielinski: Investigation. J.K. Burks: Conceptualization, resources, formal analysis, supervision, visualization, methodology. J.P. Long: Formal analysis. L.-Y. Kong: Investigation. G.N. Fuller: Conceptualization. J. DeGroot: Conceptualization. E.P. Sulman: Conceptualization, resources. A.B. Heimberger: Conceptualization, resources, supervision, funding acquisition, writing-original draft.
The authors acknowledge the Flow Cytometry and Cellular Imaging Core Facility at MD Anderson funded by the NCI No. CA16672 for their assistance with flow cytometry and imaging data acquisition, the Small Animal Imaging Facility at MD Anderson supported by the NIH/NCI under award number P30CA016672, and David M. Wildrick, Ph.D., Jennifer Everts, and Audria Patrick for their editorial and administrative support. This research was supported by the Cancer Prevention and Research Institute of Texas IIRA-RP160482, the NIH CA120813, P50 CA093459, P50 CA127001, and P30 CA016672, the Ben and Catherine Ivy Foundation, the MD Anderson GBM Moonshot, and the Brockman Foundation.
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.