Female C57BL/6 and BALB/c mice 7–8 weeks old were purchased from Taconic (Germantown, New York, USA). T cell receptor alpha knockout (TCR alpha KO) mice were a gift from Dr Marcel Wuethrich. Mice were housed in accordance with the Guide for Care and Use of Laboratory Mice. Experiments were performed under an animal protocol approved by the Institutional Animal Care and Use Committee.

B78-D14 (B78) melanoma is derived from B16 melanoma and was obtained from Ralph Reisfeld.22–24 B16-F10 melanoma was obtained from American Type Culture Collection (ATCC). B16-F10 cells were transduced to express luciferase (B16-Luc) via lentiviral transduction for in vivo imaging. The 4T1 triple-negative breast cancer cell line was obtained from ATCC. The MOC2 head and neck squamous cell carcinoma cell line was a kind gift from Dr Ravindra Uppaluri. B78, B16, 4T1, and MOC2 cells were grown in vitro as previously described.9 25–27 Mycoplasma testing via PCR was routinely done.

B78 cells (2×10⁶) and MOC2 (2×10⁶) cells were injected intradermally into the right flank of C57BL/6 mice, and 4T1 cells (2×10⁵) were injected into BALB/c mice. Tumors were measured two times per week with calipers. Tumor volumes were calculated as previously described.9 Once average tumor volumes reached the target size, mice were randomized into their treatment groups so that each group had a similar average starting tumor volume. Mice were removed from the study and euthanized when the longest tumor dimension reached 20 mm or the mouse became moribund due to its metastatic disease. Some mice that were cured of their tumor burden were rechallenged with a second inoculation (on the abdomen) of the same dose of tumor cells 90–120 days after treatment began.

For depletion experiments, anti-NK1.1 (100 μg), anti-CD4 (400 μg), anti-CD8 (400 μg), or rat IgG (400 μg) was given via intraperitoneal injection on days 1, 6, 13, 20, and 27 of the experiment. Depletion of immune subsets was confirmed as described further.

In vivo RT was dosed to flank tumors using an X-Rad 320 irradiator as previously described (Precision X-Ray).9 RT was delivered in one fraction totaling either 8 Gy (4T1, MOC2 models) or 12 Gy (B78 models). The day of RT was defined as day 0 of treatment.

BEMPEG was provided by Nektar Therapeutics. Anti-CTLA-4 (clone 9D9, IgG2c) was provided by Bristol Myers Squibb. Anti-PD-L1 (clone 10 F.G92) was purchased from Bio X Cell. IL-2 (teceleukin) was provided by the National Institutes of Health. NK 1.1 depleting antibody (clone PK136), CD4 depleting antibody (clone GK1.5), and CD8 depleting antibody (clone 2.43) were purchased from Bio X Cell. Rat IgG was purchased from Sigma Aldrich.

Intravenous injections of immunotherapy were given by retro-orbital injection. Intravenous injections of 12 μg (4T1 model) or 16 μg (B78, MOC2 models) BEMPEG dissolved in 100 μL BEMPEG buffer (buffer) were given on days 5, 14, and 23. Intravenous and IT injections of 9.375 μg IL-2 in 100 μL phosphate-buffered saline (PBS) were given on days 5–9. In experiments where BEMPEG was compared with IL-2, each cohort of mice received the IL-2 equivalent of ~750 000 U of IL-2 total per mouse. Anti-CTLA-4 (200 μg) was administered by intraperitoneal injections on days 2, 5, and 8. Anti-PD-L1 (200 μg) was administered by intraperitoneal injections on days 1, 3, and 7.

Peripheral blood (10 μL) was collected in EDTA-treated tubes from mice via facial vein on days 13 (prior to scheduled dose of depletion antibody) and 34. Red blood cell (RBC) lysis was performed with ammonium-chloride-potassium (ACK) lysis buffer (Thermo Fisher). Cells were incubated with anti-CD16/32 (Biolegend, Clone 93) for 10 min at room temperature and then stained for 30 min at 4°C with the following antibodies: CD45 BV510 (30-F11), CD4 BV785 (GK1.5), CD8 APC-R700 (53–6.7), and NK1.1 PE-CF594 (PK136) all from BD Biosciences. 4’,6’-diamindino-2-phenylindole (DAPI) was used for live/dead staining. All data were collected on an Attune flow cytometer (Thermo Fisher) and analyzed with FlowJo V.10 software (BD).

B78 tumor-bearing mice were randomized into no treatment or RT groups when tumors were ~150 mm³ (n=5 per group). On day 5 after RT, mice were euthanized, and spleens were harvested and disaggregated between two microscope slides. Cells were prepped, Fc blocked, and stained for live/dead as described previously. Each sample was stained with the following antibodies in addition to those listed previously: CD3PE-Cy5 (145–2 C11) and CD122 PE (TM-β1) all from BioLegend and CD25 BB515 (PC61) from BD Biosciences. Samples were fixed and permeated overnight at 4°C with the FoxP3/Transcription Factor Staining Buffer Set from eBioscience following kit instructions. Samples were then stained with FoxP3 PE-Cy7 (FJK-16s) from eBioscience for 30 min at 4°C. Data were collected and analyzed as described previously for blood analysis.

CD4+ T cells with Tregs (CD45+ CD4+ CD25+ FoxP3+) gated out are referred to as ‘CD4+ T-helper cells’ (acknowledging that this population can contain other non-Treg CD4+ cells).

B78 tumor-bearing mice were randomized into four treatment groups (n=5 per group). Groups were treated with buffer, BEMPEG, RT, or RT+BEMPEG as described previously. On day 14 after treatment, mice were euthanized, and tumors were excised and digested into a single-cell suspension using a Miltenyi gentleMACS Octo Dissociator (2.5 mL of complete RPMI media, 2.5 mg of collagenase, and 250 μg of DNAase per tumor). After tumor disassociation, tumor contents were passed through a 70 μm filter. In addition to those described for blood and spleen cells, the following antibodies were used: CD19 PE-Cy5 (6D5), Ly6G Alexa647 (1A8), and OX40 PE (OX-86), all from BioLegend; and Ly6C BV605 (AL-21) and CD11b V450 (M1/70), all from BD Biosciences.

Tregs and T-helper cells were gated separately within the CD4+ population.

Average group tumor volumes are plotted showing mean±SEM. Tumor volume plots were summarized by time-weighted average (area under the volume–time curve, calculated using trapezoidal method). Time-weighted averages were compared between treatment groups overall by Kruskal-Wallis tests. If significance was found using Kruskal-Wallis test, then pairwise comparisons were conducted using Mann-Whitney tests. Survival data were plotted using Kaplan-Meier methods and analyzed using log-rank comparisons. Despite the large number of tests, no p value correction methods were used to account for inflated type I error; no corrections for multiple hypothesis testing were made. Flow cytometry results are plotted as mean±SEM. Flow results were analyzed using an unpaired t-test or one-way analysis of variance analysis with Tukey’s multiple comparisons test (as clarified in the figure legends). χ² analysis was used to compare rechallenge rejection results. P values of <0.05 were considered significant and were indicated in all figures as follows: *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001; ns indicates non-significance.

Based on our previously published data showing a synergistic interaction between RT and intratumoral injectioned of immunocytokine (IT-IC),9 we hypothesized a similar interaction would occur between RT and BEMPEG, without the need for IT delivery. Mice bearing B78 tumors were randomized when average tumor volumes were 100–150 mm³ and were treated with intravenous injections of buffer, BEMPEG, local external beam RT, or RT+BEMPEG.

Compared with the vehicle control, RT alone and BEMPEG alone significantly slowed tumor progression (figure 1A) and improved survival, and achieved 0% or 20% complete regression, respectively (figure 1B). The combined RT+BEMPEG therapy enhanced treatment efficacy, with significant reductions in the tumor growth rate (figure 1A) and significantly improved survival (figure 1B). The majority of mice (67%) treated with RT+BEMPEG were cured of their tumor burden (figure 1B and online supplemental file 1), and all cured mice rejected the B78 tumor rechallenge, thus demonstrating immune memory (online supplemental figure S1B).

In the MOC2 model (average starting tumor volume ~100 mm³), BEMPEG alone and RT alone significantly slowed tumor progression compared with the vehicle control (figures 1C and online supplemental figure S1D). RT alone significantly improved overall survival (figure 1D). RT+BEMPEG significantly improved tumor growth inhibition over the three other groups and displayed a significant survival benefit over BEMPEG alone and vehicle control.

In the 4T1 model (average tumor starting tumor volume 25–75 mm³), BEMPEG alone provided minimal benefit in tumor control, while RT alone significantly slowed tumor progression. RT+BEMPEG proved to be significantly more effective in controlling tumor progression than the separate treatments (figure 1E and online supplemental figure S1C). However, RT+BEMPEG did not improve overall survival in the 4T1 model (figure 1F). The mice in all groups required euthanasia at similar times due to respiratory failure resulting from metastatic disease spread.

Since the administration of BEMPEG after RT induced antitumor activity in multiple tumor models, we hypothesized that local RT was priming the immune system for BEMPEG. To investigate this, we randomized mice bearing 100–150 mm³ B78 flank tumors and treated half with local flank RT. On day 5 after RT, spleens were excised from control and RT-treated mice.

Mice treated with RT had an increased ratio of CD4+ T-helper/CD45+ cells in the spleen and similar CD8+:CD45+ ratios as control mice (figure 2A,B and online supplemental figure S2). The ratio of CD4+ and CD8+ T cells that were copositive for CD122 significantly increased following RT (figure 2A,B). These increased ratios were accompanied by significant increases in CD122 median fluorescent intensity (MFI) on both the CD8+ T cells and CD4+ T helpers in the spleen following RT (figure 2A,B). Given BEMPEG’s preferential activity towards the CD122 subunit (the intermediate affinity, heterodimeric IL-2R), these data indicate RT is priming peripheral T cells for activation by BEMPEG.

Next, we investigated the type of immune cells that RT+BEMPEG recruits to the TME. Mice bearing 100–150 mm³ B78 flank tumors were treated with buffer, BEMPEG (day 5), RT (day 0), or RT+BEMPEG.

RT +BEMPEG significantly increased CD45+ immune cell infiltration compared with all other treatment groups, while BEMPEG and RT alone significantly increased the CD45+ immune infiltrate over the buffer control (figure 2C). There was a trend toward increased CD4+ T-helper infiltrate with RT+BEMPEG (p=0.06 vs buffer) and BEMPEG monotherapy (p=0.15 vs buffer). CD8+ T-cell infiltration was significantly greater in mice treated with RT+BEMPEG when compared with all other groups, while BEMPEG monotherapy and RT alone both demonstrated a modest increase in CD8+TIL frequency over control. Both groups treated with BEMPEG demonstrated a significant increase in NK cell tumor infiltration compared with buffer.

BEMPEG monotherapy did not significantly increase tumor Treg cell frequency over baseline. RT alone significantly increased the frequency of tumor Tregs, but RT+BEMPEG did not result in a further increase in Tregs over RT alone (figure 2C).

BEMPEG monotherapy and RT+BEMPEG both demonstrated CD4+ T-helper:Treg and CD8+:Treg ratios that were not significantly different from those for buffer. BEMPEG monotherapy significantly increased the CD8+:Treg and CD4+ T-helper:Treg ratios compared with RT alone. While RT+BEMPEG had mean CD8+:Treg and CD4+ T-helper:Treg ratios that appeared higher than RT alone, these were not significant (figure 2C).

Taken together, RT+BEMPEG increases the tumor-infiltrating CD8, T helper, and NK cell frequencies, and the increase in these populations is not accompanied by an equivalent increase in Tregs.

We conducted NK-cell and T-cell depletion experiments in the B78 model to better understand the role these cells have in the antitumor effect of RT+BEMPEG. Mice bearing ~150 mm³ B78 tumors were treated with RT, RT+BEMPEG+rat IgG, RT+BEMPEG+αNK1.1, or RT+BEMPEG+αCD4+αCD8. Depletion efficacy was confirmed via peripheral blood collection on days 13 and 34 (online supplemental figure S4).

The cohort of mice receiving RT+BEMPEG+αNK1.1 antibodies displayed similar tumor volume changes, complete response (CR) rate, and overall survival compared with the group treated with RT+BEMPEG+rat IgG (figure 3A,B). Additionally, depletion of NK cells during the treatment phase of the experiments did not negatively impact the long-term immune memory in the animals that were cured of their disease burden (online supplemental figure S5).

Mice treated with RT+BEMPEG+αCD4+αCD8 demonstrated a modest, but significant, slowing of tumor progression compared with the RT control, resulting in a significant survival benefit over RT alone (figure 3A,B) and suggesting that RT+BEMPEG is stimulating an antitumor immune effect outside of the T-cell compartment. Mice treated with RT+BEMPEG+rat IgG demonstrated significantly better tumor control and overall survival than the T-cell depletion cohort. The necessary role of T cells in the antitumor effect of RT+BEMPEG in the B78 model is supported by the lack of tumor regression following RT+BEMPEG in T cell-deficient TCR alpha KO mice (online supplemental figure S6).

Since the combination of RT and high-dose IL-2 is being tested in the clinic, we compared the antitumor strength of RT+BEMPEG to RT+IL-2 in our B78 melanoma model. Mice were randomized (average B78 starting tumor volume 100–150 mm³) and treated with RT, RT+intravenous IL-2, RT+IT IL-2, or RT+BEMPEG. Equivalent amounts of IL-2 were dosed between groups receiving IL-2 or BEMPEG.

Dosing intravenous IL-2 after RT had little effect on average tumor volume and survival (figure 4A,B). One out of 10 mice treated with RT+intravenous IL-2 completely cleared its initial tumor burden (online supplemental figure S7B). This cured mouse failed to demonstrate immune memory on rechallenge (figure 4C).

RT+IT IL-2 caused similar tumor regression and survival as RT+BEMPEG. When the results of three independent experiments were included, there was a suggestion that RT+BEMPEG could be more effective at curing animals of their disease: 16 of 21 (76%) for RT+BEMPEG vs 9 of 15 (60%) for RT+IT IL-2 (figure 4A and online supplemental figure S7A,B). However, this difference was not significant (p=0.298). Nevertheless, RT+BEMPEG did activate a stronger, long-lasting immune memory response than RT+IT IL-2. Namely, when cured mice were rechallenged with B78 melanoma, 100% of the RT+BEMPEG-cured mice rejected the rechallenge, compared with only 56% of the RT+IT IL-2-cured mice (p<0.01) (figure 4C).

Recent reports have shown that smaller baseline tumor size is a predictor of improved overall survival in patients treated with ICB.28–30 We tested RT+BEMPEG under conditions where tumors had a larger starting volume and, therefore, were more difficult to treat. Mice bearing larger B78 tumors were randomized when average tumor volumes were 400 or 1000 mm³, and each cohort received RT, BEMPEG, and/or α-CTLA-4.

In the model with 400 mm³ tumors, RT+α-CTLA-4 provided no improvement in tumor volume control or survival over RT alone. RT+BEMPEG controlled tumor progression better than RT alone or RT+α-CTLA-4, resulting in a significantly improved overall survival (figure 5A,B and online supplemental figure S8). RT+BEMPEG+α-CTLA-4 demonstrated similar tumor volume control and overall survival as RT+BEMPEG.

At a still higher B78 tumor burden of 1000 mm³ tumors (figures 5C–E and online supplemental figure S9), RT+BEMPEG slowed tumor progression compared with RT alone. Mice treated with RT+BEMPEG showed a 20% CR rate (figure 5D and online supplemental figure S9). Remarkably, RT+BEMPEG+α-CTLA-4 more than doubled the CR rate in this large tumor burden model, providing 55% (11/20) CR with improved tumor control and overall survival compared with RT+BEMPEG. The images in figure 5E highlight the striking effect of triple therapy on tumor burden, comparing day 1 vs day 66 post-treatment.

The goal of an in situ vaccine is to create a systemic tumor-specific memory response that mediates an antitumor effect at all sites of disease. To investigate the strength of RT+BEMPEG+α-CTLA-4 as an in situ vaccine, we tested the regimen in an induced metastatic model. There are limitations to preclinical models studying abscopal responses; in particular, patients have increased levels of tumor heterogeneity that are difficult to replicate in orthotopic preclinical models.31 We attempted to recapitulate tumor heterogeneity by injecting B16 cells, the parental line to B78, intravenously into mice bearing a single ~150 mm³ B78 flank tumor (primary) 1 day before mice were randomized and treated. Mice were treated with buffer or a combination of local flank RT, α-CTLA-4, and BEMPEG.

Untreated mice in the B78 primary with B16 metastasis model succumb to pulmonary compromise from their metastatic disease burden prior to reaching the flank tumor size endpoint. In this model, BEMPEG+α-CTLA-4 significantly prolonged survival compared with the buffer control (figure 6B). This effect was due to BEMPEG+α-CTLA-4 slowing the progression of lung metastasis, as BEMPEG+α-CTLA-4 failed to control flank tumor progression (figure 6A,C). RT+α-CTLA-4 improved survival over buffer, largely due to its effect on slowing the progression of metastatic disease. The majority (77%) of the mice treated with RT+α-CTLA-4 were euthanized due to primary tumor progression. RT+BEMPEG significantly improved survival over buffer by controlling both flank tumor and lung metastasis progression, although the majority of mice still required euthanasia because of metastatic lung progression (figure 6A–C). Under similar starting tumor volumes, fewer mice in the B78 primary with B16 metastasis model demonstrated complete clearance of their flank tumor following RT+BEMPEG than in the B78 primary alone model (20% vs 67% in figure 6A,B vs figure 1A,B).

Importantly, RT+BEMPEG+α-CTLA-4 clearly provided superior primary tumor control and an overall survival benefit compared with all other treatment combinations (figure 6A,B and online supplemental figure S10). Even in the context of lung metastasis, 80% of mice showed complete clearance of the flank B78 tumors, and 60% of the mice were long-term survivors (figure 6B,C). Two mice were euthanized due to complications resulting from ulcerative dermatitis (UD). We believe that this infrequent side effect of the treatment regimen may reflect a predisposition of this particular group of C57BL/6 mice; other mice in this study also developed UD but required euthanasia because of tumor burden prior to the lethal progression of UD.32

Control of lung metastasis progression by RT+BEMPEG+α-CTLA-4 was T cell mediated (figure 6D,E and online supplemental figure S13). Depletion of CD4+ and CD8+ T cells reduced the effectiveness of RT+BEMPEG+α-CTLA-4, particularly against metastatic disease; all mice that received T-cell depletion died of lung metastases.

In the 4T1 model, RT+BEMPEG and BEMPEG+α-CTLA-4 significantly slowed flank tumor progression compared with buffer, but RT+α-CTLA-4 did not (figure 7A and online supplemental figure S11). However, among these groups, only BEMPEG+α-CTLA-4 demonstrated a significant survival benefit over buffer, indicating BEMPEG+α-CTLA-4 showed some enhanced effect on metastatic progression (figure 7B). The triple combination of RT+BEMPEG+α-CTLA-4 controlled primary tumor progression and improved survival better than all other groups in the study. In a preliminary flow study on day 7 after RT, we did not observe any significant cell frequency differences in the CD8+, CD4+, Treg, myeloid-derived suppressor cell (MDSC) or macrophage populations between PBS control mice and RT+BEMPEG+CTLA-4 treated mice (data not shown). One mouse in the RT+BEMPEG+α-CTLA-4 demonstrated a CR at the primary tumor site and was a long-term survivor past 180 days. Since all the other mice in the 4T1 model were euthanized because of pulmonary complications resulting from lung metastases, the prolonged survival shown in figure 7B indicates that this triple therapy controlled the progression of spontaneous lung metastases better than all other immunotherapy combinations tested.

Finally, we evaluated the addition of α-PD-L1 to RT+BEMPEG to test whether RT+BEMPEG could be combined with different checkpoint blockade mechanisms in multiple preclinical models. Local antitumor immune effects of RT have been shown to be dependent on induction of interferon (IFN) in the TME, and PD-L1 expression in the MOC2 model has been shown to be IFN dependent.33 34 In the MOC2 head and neck squamous cell carcinoma model (HNSCC), flank tumors grew progressively when treated with α-PD-L1 alone (figure 7C,D and online supplemental figure S12). Adding BEMPEG to α-PD-L1 failed to provide additional flank tumor control or survival benefit. Adding RT to α-PD-L1 improved local tumor control and significantly increased overall survival compared with α-PD-L1 alone and BEMPEG+α-PD-L1. Importantly, RT+BEMPEG+α-PD-L1 significantly slowed flank tumor progression better than all other groups (figure 7C) and resulted in a survival benefit that was significant over all other combinations tested, except for RT+BEMPEG, where there was a trend (p=0.0604) (figure 7D).