CD73 blockade enhances the local and abscopal effects of radiotherapy in a murine rectal cancer model

Anti-mouse CD73 mAb (clone TY/23) and Rat IgG2a isotype control (clone 2A3) were purchased from BioX-Cell (West Lebanon, NH, USA). Anti-mouse mAbs for flowcytometric analysis were used as described here. FITC-conjugated anti-CD8a (53–6.7), anti-CD11b (M1/70), and PE-conjugated anti-CD4 (GK1.5), anti-CD39 (Duha59), anti-CD45 (30-F11), anti-Ly-6G/Gr-1 (RB6-8C5), anti-IFN-γ (XMG1.2), and APC-conjugated anti-CD3 (17A2), anti-CD45 (30-F11), anti-CD73 (TY/11.8), and BV421 conjugated anti-CD4 (GK1.5), anti-CD45 (30-F11) and mouse recombinant Interleukin-2 (r-IL-2) were purchased from BioLegend (San Diego, CA, USA). FcR blocking reagent was obtained from Miltenyi Biotec GmBH (Bergisch Gladbach, Germany). 7-AAD (7-Aminoactinomycin D) and FVS780 were purchased from Thermo-Fisher Scientific (Waltham, MA, USA) and BD Biosciences (Franklin Lakes, NJ, USA), respectively.

LuM-1, a highly metastatic sub-clone of murine colon cancer, colon26 [31] was kindly obtained from Dr. Oguri, Aichi Cancer Center, Japan., and maintained in DMEM supplemented with 10% FCS, 100 U/mL penicillin and 100 μg/mL streptomycin (Sigma-Aldrich, St. Louis, MO, USA). After achieving > 80% confluence, cells were removed by treatment with 0.25% (w/v) trypsin solution containing 0.04% (w/v) EDTA, and then used. The cultured cells were tested by the Mycoplasma Detection Kit (R&D Systems, Minneapolis, MN, USA) in every 3 months and cells with passages 3 to 5 were used for experiments. Female Balb/c mice age 7–8 weeks were purchased from CLEA Japan (Shizuoka, Japan) and housed in specific pathogen-free (SPF) conditions.

LuM-1 cells (1 × 10⁶) were subcutaneously injected in the right flank of 8–9 weeks-old female Balb/c mice. When the primary tumors reached a volume of 100 to 150 mm³ at day12, the mice were divided into groups with each group containing 5 ~ 8 mice to enable the statistical evaluation. Local RT was delivered using MX-160 Labo (mediXtec, Chiba, Japan), as described previously [32]. In short, anesthetized mice were held in the decubitus position, and X-ray irradiation was delivered only to the subcutaneous (sc) tumor with the remainder of the body of the mouse including the lung covered with a 5 mm lead plate. We confirmed the effectiveness of shielding by this method. Mice received 3 fractions of 4 Gy every other day (days 12, 14, 16). For immunotherapy, mice received intraperitoneal injection of 200 μg anti-CD73 mAb or Rat IgG2a isotype control on days 12, 14, 16, 19, 22 and 25. All of the mice were sacrificed with cervical dislocation on day 28, and the weight of the sc tumor and number of macroscopic metastatic nodules in lung were evaluated. All the procedures were approved by Animal Care Committee of Jichi Medical University (No 17005–02) and performed according to the Japanese Guidelines for Animal Research.

LuM-1 cells were cultured at a density of 1 × 10⁶ cells/10 cm dish and 10 Gy RT given with the MX-160 Labo and incubated for an additional 24 h. The cells were harvested, incubated with 10 μl FcR blocking reagent for 10 min at 4 °C and incubated with PE-conjugated anti-CD39 and APC-conjugated anti-CD73 mAb for 30 min at a final concentration of 2.5 μg/mL. After washing twice with staining buffer, the cells were incubated with 7-AAD for 15 min on ice and staining intensity analyzed in 7-AAD (−) live cell population using FACS Calibur (BD Bioscience, Franklin Lakes, NJ, USA). For in vivo experiments, LuM-1 (1 × 10⁶) cells were subcutaneously injected in Balb/c mice and treated with 2 fractions of 4 Gy RT as described above. Two days later, tumors were excised and digested using the Tumor Dissociation Kit, mouse (Miltenyi Biotec) with gentleMACs Dissociators (Miltenyi Biotec). After lysis of red blood cells (RBC) with RBC lysis buffer (BioLegend), cells were passed through a 40-μm filter and single cell suspensions stained with APC conjugated anti-CD73 mAb and PE conjugated anti-CD45 mAb, and the expression level of CD73 was analyzed in live tumor cell population defined in 7-AAD (−) CD45 (−) gated area.

T cells producing IFN-γ were identified by intracellular staining. Mice bearing sc LuM-1 tumors received 3 fractions of 4 Gy local RT and an intraperitoneal injection of 200 μg anti-CD73 mAb or Rat IgG2a isotype control every other day (days 12, 14, 16). The mice were sacrificed on day 18, and splenocytes (1 × 10⁶) were cultured in RPMI-1640 + 10% FCS for 6 h in the presence of 1 μl/mL brefeldin A (BioLegend) for the last 2 h. The cells were harvested, fixed, permeabilized using the fixation / permeabilization buffer (BD Bioscience) according to the manufacturer’s instructions and stained with PE-conjugated IFN-γ or isotype control and FITC-conjugated anti-CD8a, APC-conjugated anti-CD3 and BV421-conjugated anti-CD4 mAb as well as FVS780 to exclude dead cells. The ratio of IFN-γ positive cells were calculated in CD3 (+) CD4 (+) or CD3 (+) CD8a (+) gated area using LSRFortessa (BD Bioscience).

Splenocytes (5 × 10⁶) from treated mice (as described above) were cultured with 1 × 10⁶ irradiated (50 Gy) LuM-1 cells in 24-well tissue culture plates in 2 ml 10% FCS+ RPMI-1640 medium supplemented with 20 ng/ml mouse rIL-2 for 12 days. Activated splenocytes were incubated with LuM-1 at an E/T ratio of 20:1 for 4 h and all cells stained with FITC-conjugated Annexin-V (BioLegend), 7-AAD and APC conjugated anti-CD45 mAb. The ratio of 7-AAD positive dead cells was calculated in the tumor cell population defined in the FSC/SCC and CD45 (−) gated areas.

Quantitative analysis of adenosine, AMP and inosine was performed using an LC-MS system consisting of Nexera X2, LCMS-8060 and a LC/MS/MS Method Package for Primary Metabolite Version 2 (Shimadzu Corp, Kyoto, Japan) as described previously [33]. In brief, sc tumors irradiated as described above (4Gyx2) were resected at 12, 24 and 48 h after treatment and dissociated. Chromatographic separation was performed at 40 °C on a Discovery® HS F5–3 column, 150 × 2.1 mm, 3 μm, (Sigma-Aldrich) with a flow rate of 0.25 mL/min. A gradient elution of mobile phase A consisting of 0.1% of formic acid in water and mobile phase B consisting of 0.1% of formic acid in acetonitrile. The mobile phase B concentration was programmed as follows: 0% (0 min) – 0% (2.0 min) – 25% (5.0 min) – 35% (11 min) – 95% (15 min) – 95% (20 min) – 0% (20.1 min). Nitrogen gas was used as the nebulizer gas with drying gases at flow rates of 3.0 and 10 L/min, respectively. Dry air for the heating gas was at 10 L/min. Collision-induced dissociation (CID) was conducted by argon gas (purity, > 99.9995%). Interface, heat block, and desolvationline temperatures were set at 300, 400, and 250 °C, respectively. Multiple reaction monitoring (MRM) transitions for adenosine, AMP, and inosine were m/z 268.1 > 136.05, m/z 384.0 > 136.05 and m/z 269.1 > 137.05, respectively, in positive ion mode. MRM transition for 2-MES was m/z 194.0 > 80.15 in negative ion mode. The polarity switching time of the instrument was 5 ms (10 ms/cycle).

Between 2008 and 2015, 64 patients with locally advanced RC received neoadjuvant chemoradiotherapy (CRT) in the Department of Surgery, Division of Gastroenterological General and Transplant Surgery, Jichi Medical University Hospital. Patients were treated with long-course RT (a dose of 50.4 Gy in 25 fractions) using 4-field box techniques. Some patients received concurrent chemotherapy with oral UFT or S1. Radical resections were performed at 8–10 weeks after the end of CRT. The excised tumors were immediately fixed in 10% buffered formalin, and consecutive formalin-fixed paraffin-embedded 4-μm sections prepared for immunohistochemical evaluation.

After treatment with xylene and ethanol and washing with phosphate-buffered saline (PBS), tumor specimens were subjected to heat-induced antigen retrieval in citrate buffer (Muto Pure Chemicals Co., Ltd., Tokyo, Japan) followed by endogenous peroxidase blocking by Peroxidase-Blocking solution (DAKO, Santa Clara, CA, USA). The tissues were washed with PBS and incubated with 5% bovine serum albumin for 30 min to block nonspecific antibody binding. The slides were then incubated overnight at 4 °C with monoclonal antibodies against CD73 (D7F9A, Rabbit IgG, Cell Signaling Technology, Danvers, MA, USA) at a dilution of 1:200 in humid chambers overnight at 4 °C. After three 5-min washes with PBS, sections were incubated with anti-rabbit secondary antibody conjugated with peroxidase for 30 min at room temperature. After washing, the enzyme substrate 3,30-diaminobenzidine (Dako REAL EnVision Detection System, DAKO) was used for visualization and counterstained with Meyer’s hematoxylin.

Staining intensities in remnant tumor cells or stroma were independently scored from 0 to 3 (Fig. S6) by two different evaluators who were unaware of the clinical findings, and the cases were divided into high (score = 2 or > 2) and low (score < 2) expression groups by the mean score of the two evaluators. This study protocol was approved by the institutional IRB of Jichi Medical University (Rin A17–164) and conducted in accordance with the guiding principles of the Declaration of Helsinki. Written informed consent was obtained from all participants.

The expression of CD39 and CD73 in cultured LuM-1cells was examined with flow cytometry. CD39 was scarcely expressed on LuM-1 cells and not changed after RT (Fig. 1a, left panel). In comparison, LuM-1 cells positively expressed CD73 and its expression level was further enhanced 24 h after treatment with 10 Gy RT (Fig. 1a, right panel and Fig. S1A). RT (4Gy × 2) was given to sc LuM-1 tumors implanted in syngeneic mice, and CD73 expression in the LuM-1 cells recovered from resected tumors were evaluated by mean fluorescein intensity (MFI) in CD45 (−) tumor cells. Consistent with the in vitro results, CD73 expression level in LuM-1 cells was significantly enhanced compared with non-irradiated controls (Fig. 1b). MFI in CD45 (+) cells did not show significant difference (Fig. S1B).

The levels of adenosine, as well as its precursor, AMP and its metabolite, inosine, in irradiated sc tumors were examined using an LC-MS system. As evaluated by the peak height ratio against the internal standard, adenosine levels in tumors were significantly increased at 24 h after 2 cycles of RT, and inosine levels were significantly increased at 48 h after RT (Fig. 1c).

In a preliminary experiment, we confirmed that all mice developed sc tumors with micrometastases in both lungs at 12 days after subcutaneously injection of LuM-1 cells, although all mice were healthy with apparent sc tumor. When local RT (4Gy × 3) was delivered selectively to sc tumors after 12 days, the weight of the sc tumor at day 28 was significantly reduced (2.5 ± 0.61 g vs 4.8 ± 0.61 g, p < 0.05, n = 5), while the number of lung metastases was not altered. Treatment with anti-CD73 mAb alone did not show significant difference from isotype control for the sc tumor or the lung metastases (Fig. 2).

However, when RT was delivered to sc tumor together with administration of anti-CD73 mAb or isotype control, the growth of sc tumor was significantly delayed in mice treated with anti-CD73 mAb (p < 0.05, at day 18 and later) and tumor volume at day 28 was reduced to about 50% (Fig. 3a, b). Moreover, the number of lung metastases was significantly reduced in anti-CD73 mAb treated mice (1, 0 ~ 30 vs 12, 1 ~ 70, p = 0.04, n = 8). No metastases were observed in 4/8 mice treated with RT+ anti-CD73 mAb, although metastases developed in all mice in the control group (Fig. 3c, d). Same trend was observed in 2 different experiments with 2 cycles of RT although the differences were not statistically significant (Fig. S2).

We then examined lymphocyte populations in the spleens and infiltrating lymphocytes in sc tumor of tumor-bearing mice. The ratios of CD4 (+) or CD8a (+) T cells and CD11b (+) Gr-1(+) myeloid derived suppressor cells were not altered comparing the anti-CD73 mAb treated and isotype control groups (Fig. S3, S4). However, as shown in Figs. 4a and b, intracellular staining showed that IFN-γ producing cells were significantly increased in CD4 (+) and CD8a (+) T cells in anti-CD73 mAb treated mice (CD4; 10.8 ± 1.2% vs 4.7 ± 1.6%, p < 0.05, n = 6: CD8a; 16.2 ± 1.7% vs 6.9 ± 2.3%, p < 0.05, n = 6). Moreover, infiltrating lymphocytes in sc tumor showed the same trend with statistical significance in CD4 (+) population (Fig. S5).

After co-culture with irradiated LuM-1 cells and rIL-2 for 12 days, splenocytes in RT and anti-CD73 mAb treated mice tended to show increased cytotoxicity against LuM-1 with marginal significance (12.8 ± 1.5% vs 7.8 ± 2.8% at E/T ratio = 20, p = 0.053, n = 3) (Fig. 4c).

The expression of CD73 in 64 surgically resected specimens from patients with RC who had received neoadjuvant CRT was immunohistochemically evaluated. The outcomes of these patients was evaluated with regard to CD73 expression. As shown in Fig. 5, remnant cancer cells and stroma were stained positive for CD73 and the staining pattern was highly variable among the patients. Therefore, we separately evaluated the staining intensity in remnant tumor cells and stroma (Fig. S6) and divided these into high and low expression groups (Fig. 5 and Table 1). The CD73 expression level did not show significant correlation with clinical or pathological findings including pathological response (Table 1). However, recurrence in distant sites tended to be observed frequently in patients with higher-expressing CD73 tumors (Table 1). Accordingly, patients with tumors showing high CD73 expression either in remnant tumor cells or stroma tended to have shorterRFS and OS compared to patients with low CD73 expression (Fig. 6). Especially, 13 patients with tumors that highly express CD73 both in remnant tumor cells and stroma showed markedly worse outcomes compared to the other 51 patients (p = 0.0059) with mean RFS of 22 months (Fig. 6 right panels). In the univariate analysis, high CD73 expression both in remnant tumor cells and stroma was significantly associated with worse prognosis (Table S1). In the multivariate analysis, high CD73 expression both in remnant tumor cells and stroma remained an independent predictor of RFS and OS (Table S1).