Anti-inflammatory triterpenoid blocks immune suppressive function of myeloid-derived suppressor cells and improves immune response in cancer

Patients

Experiments in vitro were performed using samples of peripheral blood collected from 9 patients with histologically confirmed locally advanced or metastatic renal cell carcinoma or soft tissue sarcoma. None of the patients had been treated with chemo- or radiation therapy for at least 6 months prior to collection of blood.

Peripheral blood mononuclear cells (PBMC) from 19 patients (9 females and 10 males age 46-80) who were treated in a phase I clinical trial with RTA 402 conducted at Sammons Cancer Center (Dallas, TX) were analysed (Study ID number C-0702) . All patients were diagnosed with locally advanced (stage II-III) or metastatic (stage IV) pancreatic adenocarcinoma and were not amenable to resection with curative intent. CDDO-Me (RTA-402) was administered orally daily for 21 days. Nine patients received a dose of 150 mg/day, 2 patients - 200 mg/day, 6 patients - 250 mg/day, and 2 patients – 300 mg/day. All patients were treated with gemcitabine (1000 mg/m² i.v.) on days 1, 8 and 15 starting the week RTA-402 was initiated. Cycles were repeated every 28 days. Immunological evaluations were performed before the start of treatment and after 2 weeks of treatment with CDDO-Me. All patients provided a written informed consent in an Institutional Review Board approved protocol.

Mice and tumor models

Female C57BL/6, mice aged 6–8 week were obtained from the National Cancer Institute (Frederick, MD). Female SCID/beige mice aged 6-8 weeks of age were purchased from Taconic (Germantown, NY). Mice were kept in pathogen-free conditions and handled in accordance with the requirements of the Guideline for Animal Experiments. The following subcutaneous tumor models were used: EL-4 thymoma (obtained from American Type Culture Collection (ATCC), Manassas, VA), Lewis Lung Carcinoma (LLC), and MC38 colon carcinoma (provided by I. Turkova, University of Pittsburgh, Pittsburgh, PA). LLC-IL-1β cell line was created by transduction of pLXSN/ssIL-1β plasmid made by R. Apte (Ben Gurion University of Negev, Israel) and provided by S. Ostrand-Rosenberg (University of Maryland, Baltimore, MD) using Nucleofector kit (Amaxa). After transduction, cells were cultured for 48 hr followed by G418 selection (CalBiochem). Clones with highest IL-1β production were selected.

Isolation of cells and functional assays

MDSC were immune-magnetically isolated from spleens of tumor-bearing mice using biotinylated anti-Gr1 antibody (BD Pharmingen) and streptavidin beads (Miltenyi). The purity of Gr-1⁺CD11b⁺ cell populations was >95%. Splenocytes from OT-1 mice were used in functional tests. CD8⁺ T cells from these mice have a TCR that recognize ovalbumin (OVA)-derived peptide SIINFEKL. The number of IFN-γ producing cells in response to stimulation with 10 μg/ml specific or control (RAHYNIVTF) peptide was evaluated in an ELISPOT assay as described earlier (24). The spots were counted in triplicate and calculated using an automatic ELISPOT counter (Cellular Technology, Ltd). Peptides were purchased from American Peptide Company, Vista, CA. Cell proliferation induced by antigen-specific or CD3/CD28 antibody stimulation (0.5 μg/ml and 5 μg/ml, respectively) was evaluated using ³H-thymidine incorporation as described previously (24).

For some assays two million splenocytes were incubated with the relevant antibodies and then flow cytometry data were acquired using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA) and were analyzed using FlowJo software (Tree Star, Ashland, OR).

Dendritic cell culture and isolation of MDSC and T cells

DCs were generated from MNC obtained from healthy donor leukocyte-enriched buffy coat (Florida Blood Services) by 5-day culture with 50 ng/ml recombinant human GM-CSF (PeproTech Inc., Rocky Hill, NJ) and 6 ng/ml IL-4 (PeproTech Inc.). T cells were purified from another unrelated healthy donor buffy coat by enrichment from PBMC using human T-cell enrichment columns (R&D Systems Inc., Minneapolis, MN) according to the manufacturer instructions.

For isolation of MDSC, MNC isolated from 30 ml of patient's peripheral blood were cultured overnight in complete media at a concentration of 4×10⁶/ml. For the evaluation of phenotype cells were labeled with appropriate panels of antibodies and viable cells (DAPI⁻) were sorted into Lin⁻HLA-DR⁻CD33⁺ or CD14⁻CD33⁺CD11b^high cells using a FACSAria cell sorter (Becton Dickinson, Mountain View, CA). Lineage (Lin) markers included CD3, CD14, CD19 and CD56. All monoclonal antibodies used were from BD Pharmingen. The phenotype of the cells and post-sort purity was evaluated by multicolor flow cytometry using a FACSAria cytometer (BD Biosciences, Mountain View, CA) using FlowJo software.

Qualitative and quantitative identification of T cells secreting IFN-γ (ELISPOT assay)

The IFN-γ ELISPOT assay was performed in 96-well MultiScreen-HA plates (Millipore, Bedford, MA) coated overnight at 4°C with 1.4 μg/ml purified anti-human IFN-γ monoclonal antibody (clone 1-D1K, Mabtech, Inc., OH) in DPBS (Thermo Scientific, MA). 2×10⁵ T cells and autologous DCs were placed into each well at a 50:1 ratio in triplicates with or without MDSC at different ratios (1:2, 1:4 and 1:8). CDDO-Me was added at a concentration of 200 or 300 nM. Triplicates for positive controls (T cells + 5 μg/ml phytohemagglutinin (PHA; Sigma-Aldrich, St. Louis, MO)), and negative controls (T cells alone) were also added to the assay. Following 48 hr incubation at 37°C in a humidified 5% CO₂-incubator, plates were washed in DPBS and incubated for 2 hours with 1 μg/ml biotinylated anti-human IFN-γ monoclonal antibody (clone 7-B6-1, Mabtech, Inc., OH) followed by 1 hr incubation with 1 μg/ml streptavidin-HRP at room temperature. Spots were visualized with 50 μl/well TMB-H peroxidase substrate (Moss Inc, MD). The spots were counted with an ImmunoSpot System (Cellular Technology, Ltd, OH).

ROS production

The oxidation-sensitive dye DCFDA (Molecular Probes/Invitrogen) was used for the measurement of ROS production by MDSC. Cells were incubated at room temperature in serum-free RPMI media in the presence of 3 μM DCFDA with or without 300 nM PMA for 30 min, washed with PBS, and then labeled with anti-CD11b-PE-Cy7 and anti-Gr-1-APC antibodies. After incubation on ice for 20 min, cells were washed with PBS and analyzed using flow cytometry.

Arginase activity

Arginase activity was measured in cell lysates, as previously described (25).

NO production

Equal volumes of culture supernatants (100 μl) were mixed with Greiss reagent (1% sulfanilamide in 5% phosphoric acid and 0.1% N-1-naphthylethylenediamine dihydrochloride in double-distilled water). After 10 min incubation at room temperature, the absorbance at 550 nm was measured using a microplate plate reader (Bio-Rad). Nitrite concentrations were determined by comparing the absorbance values for the test samples to a standard curve generated by serial dilution of 0.25 mM sodium nitrite.

Dendritic cell vaccine

Bone marrow (BM) cells were obtained from the femurs and tibias of mice and were cultured in complete RPMI 1640 medium supplemented with 10% fetal bovine serum, 10 ng/lml GM-CSF, 10 ng/lml IL-4, and 50 mM 2-mercaptoethanol. On day 5, cells were collected, and DCs were enriched by centrifugation over a Nycoprep gradient (Accurate Chemical & Scientific Corporation, Westbury, NY). DCs (2×10⁶ cells) were washed and infected with adenoviruses encoding full-length survivin gene (Ad-surv) (5,000 viral particles/cell). The details of this construct were described elsewhere (26). DCs were activated with LPS (100 ng/ml) for 18 hr, washed in PBS and 5×10⁵ cells were injected s.c. into mice.

Statistical analysis

For in vitro experiments statistical analysis was performed using 2-tailed Mann-Whitney or Wilcoxon Rank Sum test with significance level of 0.05. To compare two related groups Wilcoxon Signed Rank test was used. For tumor measurements the Anderson-Darling statistics and normal curves were examined to assess normality, and if tumor measurements were normally distributed then ANOVA test was used. If the normality assumption was violated but the distributions were symmetric, Kruskal-Wallis test was used to determine whether there was significant difference in tumor sizes measured across treatment groups at each time. Tukey's method was applied for all pair-wise group comparison. All statistical analyses were performed using SAS (version 9.1; SAS Institute; Cary, NC).

Regulation of MDSC function by CDDO-Me in vitro

To assess the possible effect of triterpenoid on MDSC, Gr-1⁺CD11b⁺ cells were isolated from spleens of EL-4 tumor-bearing mice and then cultured for 24 hr in the presence of 10 ng/ml GM-CSF and different concentrations of CDDO-Me. First, we evaluated whether CDDO-Me induced up-regulation of a specific target gene in MDSC. Previous studies have demonstrated that CDDO-Me selectively up-regulated expression of NAD(P)H: quinone oxidoreductase 1 (NQO1) (27). Therefore we used NQO1 as an indicator of triterpenoid activity in these cells. CDDO-Me significantly increased NQO1 expression at concentrations starting from 30 nM (Fig. 1A). At these concentrations CDDO-Me did not affect MDSC viability. Significant cell death was observed only at a concentration of 600 nM (Fig. 1A). Several major factors are implicated in MDSC-mediated immune suppression. These include arginase, NO, and reactive oxygen species (ROS) (8). We investigated the effect of CDDO-Me on these factors. At all tested concentrations, CDDO-Me did not affect arginase activity or NO level in MDSC (Supplemental Fig. 1). In contrast, CDDO-Me significantly reduced the level of ROS in MDSC (Fig. 1B). Increased production of peroxynitrite, which is a result of interaction between ROS (superoxide) and NO is one of the hallmarks of MDSC functional activity (5, 8, 28, 29). CDDO-Me dramatically reduced the level of nitrotyrosine in MDSC, which reflects peroxynitrite activity (Fig. 1B).

Inhibition of antigen-specific CD8⁺ T cells is the main characteristic of MDSC. To assess the effect of CDDO-Me on MDSC's suppression of T-cell responses, Gr-1⁺CD11b⁺ cells were isolated from spleens of EL-4 tumor-bearing mice and cultured with OT-1 T cells in the presence of a specific peptide (SIINFEKL). MDSC significantly reduced T-cell responses to the specific peptide. CDDO-Me completely abrogated this suppressive activity (Fig. 1C).

We evaluated direct antitumor effect of CDDO-Me in vitro. In all three tested mouse tumor lines, CDDO-Me demonstrated significant antitumor activity. However, the effect was observed only at concentrations higher than 1μM (Fig. 1D). Thus, CDDO-Me, at low to midrange nM concentrations did not affect MDSC viability. However, it blocked ROS production, decreased levels of peroxynitrite, and abrogated MDSC suppressive activity against antigen-specific CD8⁺ T cells.

Effect of synthetic triterpenoid on MDSC function in tumor-bearing mice in vivo

To assess the effect of CDDO-Me on MDSC in vivo, three experimental models were used: EL-4 thymoma, LLC lung carcinoma, and MC38 colon carcinoma. All tumors were established s.c. Treatment with CDDO-Me was started on day 14 after tumor inoculation when tumor reached 7-8 mm in diameter. Mice received CDDO-Me dissolved in ethanol and Noebee oil before being mixed into powdered chow and the concentration was calculated per chow weight. We tested several doses of CDDO-Me. Toxicity was evaluated based on animal weight and gross signs of distress. The doses of 60-100 mg/kg chow did not cause toxicity during at least 14 days of continuous administration, unlike 150 mg/kg, which was only tolerated for 7-10 days.

Initially we treated mice bearing MC38 tumors at 60 mg/kg dose for 14 days and observed significant (p<0.05) decrease in tumor growth (Fig. 2A). However, when we treated EL-4 bearing mice with the same dose of CDDO-Me no effect on tumor growth was observed (data not shown). Increasing the dose of CDDO-Me to 100 mg/kg resulted in only a slight delay in tumor growth (Supplemental Figure 2) and further escalation of the dose to 150 mg/kg (given for 7 days) did not enhance the effect of the compound in this tumor model (Fig. 2A). Similar results were obtained in LLC tumor-bearing mice (Supplemental Figure 2). Analysis of the MDSC population in mice with different tumor sizes would not be very useful since tumor burden would have a direct effect on MDSC generation and function. Therefore we evaluated MDSC after one week of treatment with 150 mg/kg CDDO-Me (3 weeks after tumor inoculation) when differences in tumor sizes were not statistically significant. Triterpenoid caused a dramatic up-regulation of the target gene NQO1 and a significant decrease in the level of ROS in MDSC (Fig. 2B) without affecting arginase activity in MDSC (data not shown). All three tumor models caused significant increase in the presence of MDSC in spleens. In all these models treatment with CDDO-Me did not affect the proportion (Fig. 2C) or absolute number (data not shown) of MDSC.

To assess the effect of CDDO-Me on the presence of NT positive MDSC in vivo we injected mice with LLC cells that overexpress IL-1β. LLC-IL-1β tumor-bearing mice had dramatic expansion of MDSC in spleens (data not shown) and large presence of Gr-1⁺ cells in tumor tissues (Fig. 2D). Practically all these cells in tumor tissues were NT positive indicating production of peroxynitrite. Seven day treatment of mice with CDDO-Me resulted in significant decrease in the presence of NT positive cells. In contrast to non-treated mice tumors from CDDO-Me treated mice contained Gr-1⁺ NT⁻ cells (Fig. 2D).

Suppression of antigen-specific CD8⁺ T cells was evaluated using IFN-γ ELISPOT and proliferation assays. MDSC from all three tumor models inhibited CD8⁺ antigen-specific T-cell responses and treatment with CDDO-Me completely abrogated this suppression (Fig. 3A & B). Since the synthetic triterpenoid had significant antitumor effect in MC 38 tumor-bearing mice, we asked whether it was mediated by an immune system response. To address this question, MC38 tumors was established in immune-deficient SCID-beige mice. Mice were treated with 150 mg/kg CDDO-Me for 10 days. The antitumor effect of CDDO-Me in these mice was not statistically significant (Fig. 3C).

It was still possible that triterpenoid affected MDSC function in tumor-bearing mice indirectly by manipulating the tumor microenvironment. To evaluate the effect of triterpenoid on MDSC in vivo in the tumor-free host, OT-1 T cells were transferred i.v. into naïve, tumor-free C57BL/6 mice. MDSC were isolated from the spleens of EL-4 tumor-bearing mice and transferred into the tumor-free recipients described above 2 days later, followed by immunization with the specific peptide SIINFEKL. Recipient mice were pre-treated with CDDO-Me for 5 days prior to transfer of MDSC followed by three additional days of treatment after MDSC transfer (Supplemental Fig. 3). Eight days later lymph nodes were collected and stimulated with specific peptide. Response was evaluated by an IFN-γ ELISPOT assay. In the absence of MDSC transfer, the T cells in recipient mice demonstrated a strong antigen-specific response. Adoptive transfer of MDSC dramatically reduced this response. Treatment of mice with CDDO-Me completely abrogated this suppression (Fig. 4A).

Next, we asked whether triterpenoid might affect the immune response in tumor-bearing mice vaccinated with a tumor-associated antigen. EL-4 tumor-bearing mice were vaccinated with DC transduced with full-length survivin (26) on days 3, 10, and 17 after tumor inoculation. To maximize the effect of CDDO-Me we used several short cycles of CDDO-Me at a dose of 150 mg/kg chow. Mice were treated on days 3-7, 10-14, and 17-21 after tumor inoculation. Treatment of mice with several cycles of CDDO-Me substantially reduced tumor growth (p=0.02) (Fig. 4B). Vaccine by itself had only a moderate effect (p>0.1). Tumor growth slowed down during the first two weeks of treatment and then it resumed at the previous rate. However, addition of CDDO-Me to the cancer vaccine substantially delayed tumor progression (p=0.004) (Fig. 4B). To assess survivin-specific immune responses, T-cells were isolated from mice at the end of the treatment and re-stimulated with a survivin-derived peptide (26). Non-treated or CDDO-Me-alone-treated mice showed no specific response to the peptide (Fig. 4E). Moderate specific response was observed in mice vaccinated with DC-survivin vaccine. In contrast, mice treated with the combination of the vaccine and CDDO-Me showed significantly (p<0.05) higher antigen-specific response (Fig. 4C).

Effect of synthetic triterpenoid on MDSC in cancer patients

The data described above demonstrated that triterpenoid abrogated the suppressive activity of MDSC in mice. We asked whether a similar effect could be observed in cancer patients. We have developed an experimental system that allows for a direct evaluation of immune suppressive activity of MDSC in cancer patients. This system utilizes the ability of MDSC to present antigens in an allogeneic mixed leukocyte reaction. T cells isolated from healthy donors were used as responders. These cells were stimulated with DCs generated from unrelated healthy donors. At a DC:T cell ratio 1:50 this allogeneic system generated robust T-cell responses. MDSC were isolated from peripheral blood of cancer patients and added at different ratios to this cell mix and IFN-γ production was evaluated by ELIPOT assays. We used two previously suggested (6, 30, 31) combinations of markers to isolate these cells: Lin⁻HLA-DR⁻CD33⁺ and CD14⁻CD11b⁺CD33⁺ cells (Supplemental Fig. 4). MDSC were sorted from the blood of patients with renal cell carcinoma or soft tissue sarcoma and incubated with the DC:T cell mix (described above) at 1:2-1:8 (MDSC:T cell) ratio. For each combination of markers, blood from 3 different patients was tested. CD14⁻CD11b⁺CD33⁺ MDSC caused profound inhibition of T-cell responses to allogeneic DCs at a 1:2 ratio in all three tested patients. At a 1:4 ratio this effect was reduced but still remained statistically significant. At a 1:8 ratio the inhibitory effect was no longer observed (Fig. 5A). Lin⁻HLA-DR⁻CD33⁺ MDSC also induced profound inhibition of T cell responses at a 1:2 ratio in all three tested patients. In 2 out of 3 patients this effect was not diminished at a 1:4 ratio, and in one patient, even at a 1:8 ratio (Fig. 5B). Thus both combinations of markers were suitable for isolating human MDSC. We used the former combination to evaluate the effect of CDDO-Me in vitro. MDSC isolated from three patients with renal cell carcinoma were incubated with a mix of allogeneic DC and T cells in the presence of 200-300 nM of CDDO-Me. Triterpenoid completely abrogated the inhibitory effect of MDSC (Fig. 5C).

To evaluate the effect of CDDO-Me on MDSC and immune responses in vivo, we analyzed samples from 19 patients with pancreatic adenocarcinoma that were treated in the a phase I clinical trial RTA 402-C-0702. Patients were treated with gemcitabine (1000 mg/m²) i.v. weekly on days 1,8 and 15. CDDO-Me (RTA 402; bardoxolone methyl) was administered orally once daily for 21 days. Although treatment consisted of several 28-day cycles, immunological evaluations were performed only before the start and after 2 weeks of treatment. Such a relatively short period of evaluation was selected to minimize the possible effect of gemcitabine on MDSC and the immune system.

No toxicity attributed to CDDO-Me was observed. Treatment with RTA 402 and gemcitabine did not significantly affect the proportion of MDSC (Figs. 6A). No differences were also observed in the proportion of Lin⁻HLA-DR⁺ DCs (Fig. 6A). However, two-week treatment with CDDO-Me and gemcitabine resulted in a significant increase in the patients' T-cell responses to tetanus toxoid and PHA (Figs. 6B).