Concomitant targeting of programmed death-1 (PD-1) and CD137 improves the efficacy of radiotherapy in a mouse model of human BRAFV600-mutant melanoma

Tumors were induced on the skin of C57Bl/6J Tyr::CreER ^T2 ;Pten ^loxP/loxP ;Braf ^CA/+ mice as previously described [25, 26, 28]. In these mice, the estrogen receptor (ER) ligand tamoxifen induces expression of mutant Braf and loss of Pten in melanocytes. Briefly, 2 μl of 5-mM 4-hydroxytamoxifen (4-OHT, Sigma-Aldrich, H6278) in pure DMSO (Sigma-Aldrich, 276855) was applied topically on the flank of 4- to 8-week-old mice. Tumor outgrowth was monitored twice weekly by digital photographs of the tumor with a size reference. Tumor size was subsequently analyzed in two dimensions using ImageJ software (developed by the National Institutes of Health, USA). Mice were maintained under specific pathogen-free conditions. All mouse experiments were performed in accordance with institutional and national guidelines and were approved by the Animal Experimental Committee of the Netherlands Cancer Institute.

Rat α-mouse CD137 mAb (3H3, IgG2a) [29], derived from hybridoma culture supernatant, was protein-G purified. Rat α-mouse PD-1 mAb (RMP1-14; IgG2a) [30] was purchased from BioXCell. 2A3 mAb (BioXCell) was used as an isotype Control. Mouse α-mouse CTLA-4 mAb (9D9) was from BioXCell, and IL-2 (Proleukin) was from Novartis.

Therapy (5–10 mice per group) commenced when tumors reached ~20 mm². Radiotherapy of melanomas was conducted as described using the XRAD225-Cx system (Precision X-Ray Inc., CT, USA [22]). Briefly, mice were anesthetized with isoflurane after which a cone-beam CT scan of the mice was generated. Tumors were localized on the computed tomography (CT) scan and targeted for radiotherapy with 0.1-mm accuracy using round collimators of 1.0 or 1.5 cm in diameter. A single fraction of 14 Gy (225 peak kilovoltage (kVp), filtered with 0.3 mm of copper, 3 Gy/min) was delivered. Control mice were anesthetized and underwent a cone-beam CT scan, but were not exposed to radiotherapy.

Immunomodulatory α-PD-1, α-CD137, α-CTLA-4 or Control 2A3 mAbs diluted in PBS were administered at 100 μg/mouse intraperitoneally twice weekly for 2 weeks with the first dose delivered immediately after radiotherapy. IL-2 (in PBS) was administered as a high dose (7.2 × 10⁵ IU) twice daily for three consecutive days starting 3 days following radiotherapy. This dosing schedule was chosen to mimic the dosing schedule used in clinical trials as closely as possible [27]. Mice were killed when tumors reached 100–200 mm². A tumor size of 100 mm² was set as designated end point.

For the determination of tumor size at different time points post-treatment were the (interpolated) tumor sizes taken of all analyzable mice in each group. In case the tumor reached >100 mm² at a time point analyzed, the value was set to 100 mm². The survival curves generated represent the fraction of mice bearing tumors smaller than 100 mm². Censored events indicate mice that were killed before treated tumors reached 100 mm².

Mice bearing 2–3 established melanomas were subjected to 14 Gy localized radiotherapy to one of the tumors. One week later, single-cell suspensions were prepared from non-irradiated tumors, irradiated tumors or inguinal lymph nodes, stained with the fluorochrome-conjugated mAbs (from BD Pharmingen unless otherwise specified) indicated below and analyzed by flow cytometry according to the following gating strategy: Single-cell suspensions were gated on live (DAPI negative) cells. α-CD45-PE-Cy7 (104) was used to discriminate tumor (CD45⁻) and immune (CD45⁺) cells. NK cells, CD4 and CD8 T cells were identified using α-TCRβ-PE-Cy5 (H57-597), α-CD8-PerCpCy5.5 (53.6.7; CD8⁺ T cells: TCRβ⁺CD8⁺), α-CD4-FITC (GK1.5; CD4⁺ T cells: TCRβ⁺ CD4⁺), α-NK1.1-allophycocyanin (APC)-Cy7 (PK136; NK cells: TCRβ-NK1.1⁺). On each of these immune cell populations, CD25 (PC61), CTLA-4 (UC10-4F10-11) and PD-1 (J43; eBioscience) were detected using indicated PE-conjugated antibodies and CD137 was detected using a biotinylated antibody (17B5; eBioscience) followed by APC-conjugated streptavidin. This staining strategy allowed us to examine co-expression of CD137 and any of CD25, PD-1 and CTLA-4 in any of the CD4 T cell, CD8 T cell and NK cell subsets. Biotinylated or PE-conjugated isotype Controls were included for stainings to CD137, PD-1, CTLA-4 and CD25. The frequency of positive cells was determined by subtracting the % positive fraction in isotype Control staining from the % positive fraction in the staining specific for CD25, PD-1, CTLA-4 or CD137. This number is indicated in the graphs and was used for statistical analyses. CD137 staining was analyzed in triplicate in each sample and was averaged for statistical analysis. Numbers in text represent mean ± SEM. Samples were analyzed on a BD Fortessa. Examples of gating strategies for lymph node and tumor are presented in Supplemental Figs. 1 (lymph node) and 2 (tumor).

For immunohistochemical analysis, tumors (three mice per group) were fixed for 24 h in ethanol (50 %), acetic acid (5 %), formalin (3.7 %), embedded in paraffin, randomly sectioned at 4 µm. Staining was performed as previously described [31]. Briefly, fixed sections were rehydrated and incubated with primary antibodies. Endogenous peroxidases were blocked with 3 % H₂O₂ and stained with biotin-conjugated secondary antibodies, followed by incubation with HRP-conjugated streptavidin–biotin complex (DAKO). Substrate was developed with either 3-amino-9-ethylcarbazole (AEC) or diaminobenzidine (DAB) (DAKO). Primary antibodies were α-CD3 (clone SP7, cat. RM-9107 Thermo Scientific), α-CD4 (cat. 14-9766 eBioscience), α-FoxP3 (cat. 14-5773 eBioscience).

CD8 staining was performed on optimal cutting temperature compound (OCT) embedded, cryopreserved tumor pieces using standard procedures. Briefly, tumor pieces were thawed to room temperature, rehydrated in PBS and blocked for avidin and biotin (Vector SP-2001). After sections were blocked in 5 % normal goat serum and 2.5 % BSA, sections were incubated for 1 h with primary α-CD8 antibody (clone 2.43). After washing, sections were incubated with biotinylated secondary antibodies, followed by incubation with HRP-conjugated streptavidin–biotin complex and substrate was developed with DAB.

Slides were counterstained with hematoxylin and slides scanned using the Aperio ScanScope (Leika) (20× objective). ImageJ software was used to quantify # positive cells (CD3, CD4, FoxP3) or % positive area (CD8) from 3 to 5 random fields of view (FOV) per slide.

Statistical differences between groups were analyzed with the Mann–Whitney U test using GraphPad Prism (GraphPad Software) and considered significant when p < 0.05.

First, we investigated whether relevant cell surface receptors were available for targeting on melanoma tumor-infiltrating lymphocytes (TILs) before and after radiotherapy. For this purpose, CD4⁺ and CD8+ T cells and natural killer (NK) cells were examined for expression of CD25 (IL-2 receptor α-chain), CTLA-4, PD-1 and CD137. Examination of CD25 was chosen because of potent combined effects of SBRT and IL-2 in the clinic [27]; CTLA-4 and PD-1 because of potent (combined) efficacy of α-CTLA4 and α-PD-1 mAbs in late-stage melanoma patients [32] and CD137 because of potent combined effects of α-CD137/α-PD-1 mAbs and SBRT in mouse breast cancer models [14, 22].

In mice bearing 2–3 melanomas, one of these tumors was subjected to 14 Gy SBRT. Pilot experiments revealed that this radiotherapy dose induced tumor growth delay of irradiated tumors without inducing complete tumor regression. Therefore, this dose provided a window to read out the combined effect of immune-modulatory agents on radiotherapy-induced tumor growth delay.

One week after radiotherapy, single-cell suspensions were prepared from irradiated and non-irradiated tumors of the same mice, as well as from their (inguinal) lymph nodes and flow cytometric detection of receptors was performed (See Supplemental Figures 1 and 2 for gating).

CD25 was expressed in non-irradiated tumors (Tu-mock) on the majority of CD4⁺ T cells (60.8 ± 2.6 %), on small populations of CD8⁺ T cells (6.0 ± 2.7 %) and NK cells (6.1 ± 5.6 %). Radiotherapy did not significantly alter this expression. In lymph nodes, CD25 was expressed on a subset of CD4⁺ T cells (17.9 ± 2.4 %) and on a small proportion of NK cells (4.8 ± 3.7 %) and was hardly found on CD8⁺ T cells (1.5 ± 1.4 %; Fig. 1a).

CTLA-4 was expressed in non-irradiated tumors on a proportion of CD4⁺ T cells (7.4 ± %1.5 %), a subset of NK cells (4.2 ± 3.1 %), but not on CD8⁺ T cells. In irradiated tumors, CTLA-4 was occasionally detected on CD8⁺ T cells (2.5 ± 4.9 %). In lymph nodes, CTLA-4 was only expressed on a small percentage (1.2 ± 0.6 %) of CD4⁺ T cells (Fig. 1b).

PD-1 was expressed in non-irradiated tumors on the majority of both CD4⁺ and CD8⁺ T cells (means 53.0–78.1 %) and on NK cells (25.6 ± 7.3 %). Radiotherapy did not significantly alter PD-1 expression. Expression of PD-1 in lymph nodes was found to a lesser extent as compared to TILs (CD4⁺ T cells: 12.4 ± 4.6 %, CD8⁺ T cells: 2.5 ± 1.5 %, NK cells: 2.6 ± 1.3 %, Fig. 2a).

CD137 was detected in non-irradiated tumors on CD4⁺ T cells (10.17 ± 2.2 %), a small fraction of CD8⁺ T cells (1.5 ± 0.8 %) and a sizable fraction of NK cells (26.5 ± 2.5 %). Radiotherapy slightly increased the frequency of CD137-expressing CD4⁺ and CD8⁺ T cells, but this did not reach statistical significance. In lymph nodes, CD137 expression was detected on a fraction of NK cells (5.9 ± 1.4 %), but expression on CD4⁺ and CD8⁺ T cells was negligible (Fig. 2b).

Similar data were obtained when TILs were analyzed 2 days after radiotherapy (Supplemental Figure 3). Due to the small number of T cells recovered from these tumors, the variability within the groups is relatively large, leaving us unable to draw strong conclusions. However, we again observed no significant differences between TILs in irradiated versus non-irradiated tumors.

Foxp3, the hallmark protein of regulatory T cells (Tregs) was detected in 30–75 % of the intratumoral CD4⁺ T cells. Their frequency was not significantly altered by radiotherapy (Supplemental Figure 4a). In addition, radiotherapy did not alter the frequency of CD4⁺ T cells, CD8⁺ T cells and NK cells (Supplemental Figure 4b). CTLA-4, CD25 and CD137 were expressed on a higher frequency of regulatory (Foxp3+) CD4 TILs cells than on ‘non-regulatory’ (Foxp3−) CD4 TILs cells, whereas PD-1 expression was similar between the two subsets. This was not significantly affected by radiotherapy (Supplemental Figure 5).

The presence of CD25, CTLA-4, PD-1 and/or CD137 on T cells and NK cells indicates that IL-2 or antibodies targeting these receptors may have an immune modulatory and potentially therapeutic effect. We therefore tested the effectiveness of IL-2 or mAbs targeting CTLA-4, PD-1 and CD137 alone and in combination with SBRT.

Mice bearing established (>20 mm²) tumors were divided into the following treatment groups that consisted of immunotherapy alone or in combination with 14 Gray (Gy) SBRT: (1) Control Ig (Ctr), (2) IL-2, (3) α-PD-1 and α-CTLA-4, (4) α-PD-1, (5) α-CD137, (6) α-PD-1 and α-CD137, (7) SBRT, (8) SBRT + IL-2, (9) SBRT + α-PD-1 and α-CTLA-4, (10) SBRT + α-PD-1 and α-CD137, (11) SBRT + α-PD-1 and (12) SBRT + α-CD137.

Treatment in all groups was well tolerated, and adverse events such as weight loss or diarrhea were not observed. None of the immunotherapeutic approaches exhibited any single-agent activity (Fig. 3). Tumor growth and mean tumor doubling times (TDTs) in these groups (IL-2: 9.7 days, α-PD-1/α-CTLA-4: 6 days, α-PD-1: 8.7 days, α-CD137: 11.7 days, α-PD-1/α-CD137: 11 days) were statistically not significant from Control-treated mice that had a mean TDT of 9.2 days (Fig. 3). SBRT-induced significant tumor growth delay (Fig. 3, 4a) and TDT increased from 9.2 days (Control) to 40.8 days (SBRT, p < 0.0001). IL-2 or α-PD-1/α-CTLA-4 treatment did not enhance the anti-tumor effect of SBRT (mean TDT of 40.7 (p = 0.71) and 36.8 days (p = 0.60), respectively). However, concomitant targeting of PD-1 and CD137 significantly enhanced the therapeutic effect of SBRT (mean TDT of 65.1 days compared to 40.8 days of SBRT, (p = 0.0006)). α-PD-1 or α-CD137 mAbs alone did not significantly enhance the anti-tumor effect of SBRT (mean TDT: SBRT + α-PD-1: 32.3 days, SBRT + α-CD137: 43.1 days vs. SBRT: 40.8 days).

Analysis of tumor sizes at different time points after initiation of therapy (day 9, day 35 and day 55) revealed an immediate effect of SBRT on tumor growth delay: At day 9, tumor sizes of all groups that received SBRT were significantly smaller compared to mock-treated mice, while immunotherapy again did not reveal any single-agent activity. At day 35, when most of tumors of non-irradiated mice had grown out, there was a statistically significant difference in tumor size between mice that received SBRT and mice that received SBRT + IL-2, SBRT + α-CTLA-4/α-CD137, SBRT + α-PD-1/α-CD137 or SBRT + α-CD137. At day 55, a time point at which most tumors treated with SBRT had started to grow out; significant differences in tumor sizes were still observed between mice treated with SBRT and mice treated with SBRT + α-PD-1/α-CD137 or SBRT + α-CD137 (Fig. 4b).

Finally, survival curves that were generated from these data also corroborated our findings: None of the immunotherapeutic approaches alone significantly extended survival compared with mock-treated mice. SBRT significantly extended survival compared to mock-treated mice (median survival SBRT: 54 days, mock treatment: 16 days, p < 0.0001). Concomitant targeting of PD-1 and CD137 significantly extended survival of mice that were treated with radiotherapy only (median survival SBRT: 54 days, SBRT + α-PD-1/α-CD137: 72 days, p = 0.0005). In this analysis, α-CD137 treatment also slightly extended survival compared to mice that were treated with radiotherapy only (median survival SBRT: 54 days, SBRT + α-CD137: 66 days, p = 0.02), while α-PD-1 treatment had no such effect (median survival SBRT + α-PD-1: 49 days, NS compared with SBRT only; Fig. 4c).

Next, we aimed to identify the immune cell subsets that contributed to the therapeutic effect of SBRT in combination with α-PD-1 and α-CD137 mAbs. We first analyzed tumors for immune cell infiltrates by immunohistochemistry at day 31 post-start treatment, a time point at which tumors were still in a stable phase (Fig. 3, bottom right panel). Radio-immunotherapy resulted in increased frequencies of CD4⁺ and CD8⁺ T cells in the melanomas (88-fold for CD4⁺ T cells and 60-fold for CD8⁺ T cells, Fig. 5). This effect was specific to treatment by radio-immunotherapy, since mock treatment, radiotherapy or immunotherapy alone did not increase T cell frequencies at day 31 post-start treatment. In addition, the frequency of cells expressing Granzyme B (a cytolytic effector molecule found in the granules of cytotoxic T cells and NK cells) was also significantly increased in mice treated with radio-immunotherapy compared to all other groups (Supplemental Figure 6). Finally, we observed a trend in that effector quality, on a per-T cell basis was increased in mice treated with α-CD137/α-PD-1 therapy alone or in combination with radiotherapy. This was revealed by a higher frequency of T cells expressing CD43 and producing TNF-α following PMA/ionomycin stimulation following treatment with α-CD137/α-PD-1 therapy (Supplemental Figure 7). Together, these data suggest that CD4⁺ and/or CD8⁺ T cells may contribute to controlling melanoma outgrowth by radio-immunotherapy.

To test this possibility, we depleted CD4⁺ and/or CD8⁺ T cells in melanoma-bearing mice just before initiating radio-immunotherapy (Fig. 6a). Combined depletion of CD4⁺ and CD8⁺ T cells reduced the therapeutic effect of radio-immunotherapy when compared to mock depletion (50.7 mm² versus 34.7 mm² for CD4/CD8 depletion, p = 0.01). In addition, despite not reaching statistical significance, the TDT of CD4⁺ and CD8⁺ T cell-depleted mice treated with radio-immunotherapy (32.3 days) was shorter compared to mock-depleted mice (43.6 days, p = 0.25 Fig. 6b–d). Depletion of NK cells did not alter the therapeutic effect of radio-immunotherapy (27.6 mm² versus 34.7 mm² for NK depletion versus mock depletion, p = 0.20).

Collectively, our data suggest that concomitant triggering of CD137 and blocking of PD-1 signalling within irradiated melanomas enhance the intratumoral presence of both CD4⁺ and CD8⁺ T cells, which are in part required for melanoma Control.