Anti–PD-1 and Extended Half-life IL2 Synergize for Treatment of Murine Glioblastoma Independent of Host MHC Class I Expression

Glioblastoma (GBM) is a highly fatal cancer with near universal recurrence (1), even after aggressive resection (2). While combination chemotherapy and radiotherapy, biologics like bevacizumab (3), and other approaches such as tumor-treating fields (4) have all improved overall survival these increases have only been incremental (3, 5). Therapies that establish durable remissions remain elusive. The immune-specialized status of the central nervous system (CNS) complicates GBM therapy, with restricted immune cell entry and high levels of immune-dampening molecules like TGFβ, IDO1, and programmed death-ligand 1 (PD-L1) that suppress immune responses against autochthonous tumors (6). Immunosuppression exerted by intracranial tumors on peripheral adaptive immune responses further complicates immunotherapy. Reductions in circulating T-cell numbers and functionality, reduced thymic and splenic cellularity, and T-cell sequestration in bone marrow are all systemic immunosuppressive hallmarks observed in patients with GBM and experimental models (7–9).

Characteristics intrinsic to individual GBM tumors also complicate their treatment. GBM tumors have extensive mutational heterogeneity. The best characterized GBM tumor-specific antigen, EGFRvIII, is only present in approximately 30% of primary tumors (10). GBM tumors also undergo extensive intra-tumoral immune editing, with significant alteration of expressed antigens between the initial and the recurrent tumor (10–12). These aspects of GBM complicate the use of antigen-targeted vaccines. Broader immunomodulatory agents, including antibody blockade of CTLA-4 or programmed cell death protein 1 (PD-1), are agnostic to particular tumor antigens and can, consequently, be given to a wider patient population (13). However, these two interventions as single agents have failed to demonstrate efficacy over standard of care in patients with GBM (13, 14). These challenges are recapitulated in the GL261 and CT2A murine models of GBM, which are characterized as being resistant to immune checkpoint blockade (ICB) monotherapy in advanced stage of disease (15–17). For these reasons, combination therapy strategies which modulate multiple immune pathways are needed to overcome the challenges confronting GBM immunotherapy.

The FDA approved IL2 therapy for metastatic renal cell carcinoma and metastatic melanoma in 1992 and 1998, respectively (18). High doses were required to be efficacious, yet these high doses carried the risk of significant morbidity largely associated with vascular leak syndrome (19). The short half-life (t_1/2) of IL2 in circulation is a significant barrier in establishing a nontoxic dosing strategy, necessitating large doses to keep serum levels consistently within a therapeutic range (19). One engineering approach to overcome this challenge is the generation of a recombinant IL2 covalently linked to mouse serum albumin (MSA-IL2; ref. 20). MSA-IL2 has been shown to have extended t_1/2in vivo over recombinant wild-type (WT) IL2, and to have antitumor activity as part of a combination immunotherapy strategy validated in multiple tumor models (20, 21). MSA-IL2 complements the off-the-shelf appeal of anti–PD-1 ICB while acting upon independent pathways to promote antitumor activity. Following the success of extended-half-life (t_1/2) IL2 molecules in combination immunotherapies, we now successfully demonstrate efficacy with ECB against the GL261 model of GBM, a profoundly immunosuppressive tumor model when therapy is administered at late time points as in the current study (7, 20–23). Herein, we detail the treatment of established GL261 tumors by the supplementation of traditional anti–PD-1 ICB by concomitant administration of MSA-IL2. This enhanced checkpoint blockade (ECB) strategy clears established GL261 gliomas that are resistant to either MSA-IL2 or anti–PD-1 ICB alone. Most strikingly, ECB is equally efficacious in MHC class I–deficient animals as in WT animals, suggesting that the identification of patients with particular MHC Class I–restricted tumor antigens might be nonessential for future translation of this therapy.

Female WT C57BL/6 mice were purchased from Jackson Laboratory (Catalog no. 000664, Bar Harbor, Maine). The generation of H-2K^b LoxP parental mice and the MHC class I–deficient CMV-Cre parental mice are detailed in our previous work (24). Genotypes were confirmed by PCR for Cre-Recombinase using primers from the Jackson Laboratory. Flow cytometry for surface H-2K^b MHC class I surface protein ensured the appropriate identity of all experimental animals. Mice were bred and maintained in the animal facility at Mayo Clinic under Institutional Animal Care and Use Committee (IACUC) guidelines. Animal handling and all procedures were done in the manner approved by the Mayo Clinic IACUC.

GL261 cells expressing Luciferase (GL261-Luc) were the generous gift of Dr. John Ohlfest (Masonic Cancer Center, University of Minnesota, Minneapolis, MN). CT2A cells were provided by the laboratory of Dr. Richard Vile (Mayo Clinic, Rochester, MN). Both cell types were grown in DMEM (Gibco, Gaithersburg, MD) with l-glutamine media supplemented with 10% FBS and 1% penicillin/streptomycin (Sigma, St. Louis, MO). TrypLE Express was used to detach confluent cells (catalog no. 12605–010 Gibco, Gaithersburg, MD) before they were washed and counted.

Our protocol for tumor inoculation has been previously published (7). In short, mice were anesthetized by intraperitoneal injection of 100 mg/kg ketamine and 10 mg/kg xylazine. Once under anesthesia, the top of the mouse's head was cleaned with Betadine and a 0.5-cm incision was made using a sterile scalpel. The skin was then separated at the incision site and a hole was drilled into the right frontal side of the skull about 1-mm lateral and 2-mm anterior to bregma. Using a stereotactic frame, the needle of a Hamilton syringe (Hamilton Company, Reno, Nevada) was positioned in the drilled hole and lowered 3.3 mm into the cortex and retracted 0.3 mm. 60,000 GL261-Luc cells or 10,000 CT2A cells were injected in 1.5 to 2 μL total volume PBS into the pocket created in the 0.3-mm retraction site. We injected half of the volume followed by a 2-minute pause. The second half of the volume was then injected followed by a 3-minute pause after which the needle was slowly removed. The wound was sutured with Ethicon 4–0 vicryl (Ethicon Inc, Somerville, NJ) suture. WT mice were 6 to 12 weeks old at time of intracranial injection unless otherwise specified, while H-2K^b LoxP mice and CMV-Cre x H-2K^b LoxP mice were 6 to 12 months old. For that experiment, age matched WT mice were acquired.

Tumor burden in GL261-Luc–bearing animals was assessed using bioluminescence imaging as previously described (24, 25). Animals were intraperitoneally injected with 15 mg/kg D-luciferin sodium salt in PBS (Gold Biotechnology, Olivette, MO and, 10-minute later, anesthetized with 2.5% isoflurane before imaging. Anesthesia was maintained during imaging with 0.5% to 1% isoflurane. Animals were scanned using Mayo Clinic's IVIS Spectrum system (Xenogen Corp., Amameda, CA) running Living Image software. Bioluminescence intensities above 10⁵ photons/second (p/s) were considered indicative of a tumor while intensities below 10⁵ p/s fell below background levels.

As described in previous publications from our group (25, 26), we performed T2-weighted MRI imaging with a Bruker DRX-300 (300 MHz 1H) 7 Tesla vertical-bore small animal imaging system (Bruker Biospin, Billerica, MA). Mice were maintained under anesthesia during imaging by inhalation of 3% to 4% Isofluorane and their respiration was monitored for the duration of the scan. The scan used a RARE pulse sequence with repetition time (TR) = 1500 ms, echo time (TE) = 70 ms, RARE factor: 16, field of view (FOV): 3.2×1.92×1.92 cm, matrix: 256×128×128. Volumes of the tumors were determined using Analyze 12.0 software (Biomedical Imaging Resource, Mayo Clinic, Rochester, MN) as traced by a blinded reviewer on the transverse plane and then generated into a three-dimensional render.

GL261-Luc–bearing mice were imaged by the Xenogen IVIS system on D14 post inoculation and matched into treatment arms by tumor bioluminescent output. CT2A-bearing mice were imaged by T2-Weighted MRI on D12 post inoculation and matched into treatment arms by slice of maximal tumor area. In this way, groups were created with equivalent tumor burden. Mice received intraperitoneal injections of 50-μL PBS (Corning, Corning, NY) + 50 μL of 4 mg/mL IgG2a Isotype Control (Clone 2A3, catalog no. BE0089, BioXCell, Lebanon, NH), 50-μL PBS plus 50 μL of 4 mg/mL anti–PD-1 (Clone RMP1–14, catalog no. BE0146, BioXCell, Lebanon, NH), 50-μL PBS plus 50 μL of 0.6 mg/mL MSA-IL2, or 50 μL of each the anti–PD-1 and MSA-IL2 solution. MSA-IL2 (MIT, Boston, MA) was synthesized in the manner previously described (21).

Spleens were weighed before dissociation using the rubber end of a 3-mL syringe in RPMI1640 (Gibco, Gaithersburg, MD). The Spleen/RPMI solution was centrifuged at 400 × g for 10 minutes and the supernatant was discarded. The pellet was resuspended in 1 mL ACK lysis buffer (8.3 g ammonium chloride, 1 g potassium chloride, 250 μL of 0.5 mol/L EDTA made to 1,000 mL with water pH 7.2–7.4) for 3 minutes and quenched with PBS (Corning, Corning, NY) to remove red blood cells. Following multiple washes, the remaining cells were counted on a hemocytometer (Hausser Scientific, Horsham, PA) using trypan blue exclusion (Gibco, Gaithersburg, MD). Thymi were processed in an analogous manner, but were dissociated immediately after dissection. Perfused brains were mechanically homogenized using the dounce method followed by 30% percoll (Sigma, St. Louis, Mo) centrifugation as previously published (27). Recovered cells were counted as described above and stained with appropriate antibodies for 30 minutes at 4°C, washed twice with PBS (Corning, Corning, NY), and analyzed by an LSR II (BD, Franklin Lakes, NJ) or Cytek Aurora (Cytek Biosciences, Fremont, CA).

Stock solution of 1,000 unit heparin (Sigma, St. Louis, MO) in water was diluted 1:4 with PBS for a working solution. Twenty to 80 μL blood was collected from a cut made over the tail vein of restrained mice and transferred to the heparin working solution with a small transfer pipette (Samco Scientific, San Diego, CA, catalog no. 231). The blood/heparin solution was centrifuged at 400 × g for 5 minutes and the supernatant was discarded. The pellet was resuspended in 2 mL ACK lysis buffer (8.3 g ammonium chloride, 1 g potassium chloride, 250 μL of 0.5 mol/L EDTA made to 1,000 mL with water pH 7.2–7.4) for 3 minutes and quenched with PBS (Corning, NY). After two washes, pellets were stained with a panel of antibodies for 30 3 minutes (4°C), washed one additional time, and then analyzed by an LSR II flow cytometer (BD, Franklin Lakes, NJ).

PerCP Ly6C (Clone HK1.4, BioLegend, catalog no. 128027), BV650 CD279 (Clone RMP1–30, BD Biosciences, catalog no. 748266), PE CD122 (Clone TM-β1, BioLegend, catalog no. 123210), AF700 CD25 (Clone PC61, BioLegend, catalog no. 102024), APC/FIRE 810 B220 (Clone RA3–6B2, BioLegend, catalog no. 103278) were used at 1:100. PE-Cy5 CD11b (Clone M1/70, Tonbo, catalog no. 55–0112-U100) and Spark Blue 550 CD8 (Clone 53–6.7, BioLegend, catalog no. 100780) were used at 1:200. Each of BV510 F4/80 (Clone BM8, BioLegend, catalog no. 123135), BV510 CD4 (Clone RM4–5, BioLegend, catalog no. 100559), BV570 CD8α (Clone 53–6.7, BioLegend, catalog no. 100740), and BV711 Ly6G (Clone 1A8, BioLegend, catalog no. 127643) were used at 1:250 while both BV421 TCRβ (Clone 57–597, BioLegend, catalog no. 109229) and PerCP CD4 (Clone GK1.5, BioLegend, catalog no. 100432) were used at 1:400. APC CD62 L (Clone MEL-14, Tonbo, catalog no. 20–0621-U100), APC L6G (Clone 1A8, BioLegend, catalog no. 127614), APC TER119 (Clone TER-119, Tonbo, catalog no. 20–5921-U025), APC GR1 (Clone RB6–8C5, BioLegend, catalog no. 108412), APC CD11b (Clone M1/70, BioLegend, catalog no. 101212), APC-FIRE750 TCRβ (Clone H57–597, BioLegend, catalog no. 109246), APC-FIRE750 CD115 (Clone APS98, BioLegend, catalog no. 135535), BB115 CD11b (Clone M1/70, BD Biosciences, catalog no. 564454), BV421 MHC Class II Clone M5/114.15.2, BioLegend, catalog no. 107632), BV421 CD150 (Clone Q38–480, BD Biosciences, catalog no. 562811), BV785 CD8α (Clone 53–6.7, BioLegend, catalog no.100750), FITC CD103 (Clone M290, BD Biosciences, catalog no. 557494), FITC NK1.1 (Clone PK136, Tonbo, catalog no. 35–5941-U500), PacBlue MHC Class II (Clone M5/114.15.2, BioLegend, catalog no. 107620), PacBlue CD44 (Clone IM7, BioLegend, catalog no. 103020), PE FoxP3 (Clone 3G3, Tonbo, catalog no. 50–5773-U100), PE CD34 (Clone RAM34, BD Biosciences, catalog no. 551387), PE CD4 (Clone GK1.5, BD Biosciences, catalog no. 553730), PE-Cy7 TCRβ (Clone H57–597, BioLegend, catalog no.109221), PE-Cy7 CD48 (Clone HM48–1, BD Biosciences, catalog no. 560731), PerCP CD45 (Clone 30-F11, BioLegend, catalog no. 103130), and Fc Block (Clone 2.4G2, BD Biosciences, catalog no. 553141) were all used at 1:500 dilution. APC-FIRE750 CD62 L (Clone MEL-14, BioLegend, catalog no. 104450), BV650 CD44 (Clone IM7, BD Biosciences, catalog no. 740455), BV650 Sca1 (Clone D7, BD Biosciences, catalog no. 740450), BV650 CD69 (Clone H1.2F3, BD Biosciences, catalog no. 740460), BV786 CD4 (Clone RM4–5, BD Biosciences, catalog no. 563727), BV786 c-kit (Clone 2B8, BD Biosciences, catalog no. 564012), PE Ly6C (Clone HK1.4, BioLegend, catalog no. 128007), PE-CF594 CD45 (Clone 30-F11, BD Biosciences, catalog no. 562420), PE-CF594 CD62 L (Clone MEL-14, BD BioSciences, catalog no. 562404), SparkNIR B220 (Clone RA3–6B2, BioLegend, catalog no. 103268), ZombieUV Dye (BioLegend, Part: 77474) were all used at 1:1,000 dilution. Single cell suspensions were analyzed using a Cytek Aurora (Cytek Biosciences, Fremont, CA). Data were analyzed on FlowJo 10.1 (FlowJo LLC, Ashland, OR).

Anti-mouse depleting antibody targeted towards CD4 (Clone GK1.5, catalog no. BE0003–1, BioXCell, Lebanon, NH) or anti-mouse depleting antibody targeted towards CD8 (Clone YTS169.4, catalog no. BE0117 & Clone 53–6.7, catalog no. BE0004–1) was injected on days 10, 11, and 12 post GL261-Luc inoculation. Intraperitoneal administrations of 200 μg at each administration continued every 5th day afterward up until each animal had reached IACUC-approved endpoint.

All data presented as mean ± SEM or as percent survival. All analyses were conducted with GraphPad Prism 7.0 (La Jolla, CA). Tests used include a two-sided Student t test, a one-way ANOVA with Dunn multiple comparisons test, and a Mann–Whitney Rank Sum Test if the data did not follow a normal distribution. Survival analyses conducted with Mantel–Cox tests run between Control-treated WT animals and each other therapeutic group. Multiple comparisons for survival were assessed for significance with the Bonferroni correction. For all statistical analyses: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

All animal studies were approved by the Mayo Clinic's IACUC, including the animal breeding, tumor implants, tumor rechallenge, immunotherapy administration, longitudinal imaging and bleeds, T-cell depletions, and euthanasia for organ harvest. All these procedures are on our protocol for both the CT2A and GL261 glioma models.

The data generated in this study are available upon request from the corresponding author.

Enhanced t_1/2 MSA-IL2 has shown efficacy in previous combination immunotherapies in preclinical models of non-CNS cancers (20–23). We hypothesized that MSA-IL2 could enhance efficacy of anti–PD-1 ICB against the GL261 glioma model. Therefore, we inoculated mice intracranially with 6×10⁴ luciferase-expressing GL261 cells (GL261-Luc). Untreated mice bearing orthotopic GL261 gliomas have a median survival time of 30 days post inoculation (25, 26, 28–30). We began all treatment regimens at day 14 (D14) post-tumor inoculation, when tumors are established and readily detectable by MRI (ref. 29; Fig. 1A). This relatively late time point in the GL261 model more closely recapitulates human disease in which established tumors form the basis of diagnosis and treatment (31). We sorted mice into treatment arms on D14 based on the bioluminescent intensity of their GL261-Luc tumor, equalizing tumor sizes among groups (Fig. 1B). MSA-IL2 or anti–PD-1 monotherapies offered minimal survival improvement compared with PBS sham therapy (Fig. 1C). In contrast, approximately 50% of animals treated with ECB combination therapy demonstrated survived long-term, with a median survival of 57 days post tumor inoculation (Fig. 1C). Because weight loss is a commonly measured proxy for toxicity during immunotherapy administration for mice (32–34), we tracked weights over the course of our experiment. Mouse weights were not significantly different between groups on the day of tumor implant (Fig. 1D). Over the first 26 days of tumor growth, ECB-treated mice failed to lose weight in a manner that would be consistent with toxicity, and instead increased their weight relative to both the Control and αPD-1 treated groups (Fig. 1E). The rapid clearance of established GL261-Luc gliomas can be appreciated on a per-animal basis by weekly bioluminescent tracking of tumor growth (Fig. 1F–I). The majority of animals receiving ECB whose tumor bioluminescent signature falls below the limit of detection at 1×10⁵ p/s have done so by D21, even prior to the second dose of therapy (Fig. 1F). These observations on the efficacy of ECB hold true across multiple experiments pooled together (Fig. 1J–N).

To determine if ECB confers long-lasting antitumor immunity, we next inoculated a cohort of mice with GL261-Luc on D0 and treated these animals with ECB on D14 and D21 to generate a cohort of ECB Responders. This cohort of tumor-experienced long-term survivors, along with a group of age-matched tumor-naïve animals, received GL261-Luc tumor implants on D75 (Fig. 1O). Neither the tumor-naïve nor tumor-experienced group received therapeutic intervention following tumor inoculation, but only the tumor-naïve animals developed detectable tumor bioluminescent signatures by D89, 14 days following the rechallenge (Fig. 1P). All tumor-experienced mice successfully cleared a secondary challenge with GL261-Luc without need for further treatment. Meanwhile, no naïve animals survived inoculation with GL261-Luc (Fig. 1Q).

We next sought to determine the state of the peripheral immune system in GL261-Luc–bearing mice treated with ECB, as unchecked glioma growth is now understood to be an innately immunosuppressive event. Prior publications from our laboratory and from others have shown that glioma growth promotes involution of the thymus and secondary lymphoid organs, the sequestration of T cells into the bone marrow, and the reduction in MHC Class II expression on peripheral antigen presenting cells (7–9). In GL261-Luc–bearing mice treated with ECB combination therapy, the peripheral blood was analyzed by flow cytometry at day 26 posttumor inoculation, as displayed in the accompanying uniform manifold approximation and projection (UMAP) with a heat map for MHC Class II expression (Fig. 2A). We observed no statistically significant increase in T-cell counts at D26, but ECB treatment did promote a relative increase in the abundance of activated T cells (Fig. 2B–E) as well as increased MHC Class II expression among B cells and monocytes to PBS-treated glioma-bearing animals (Fig. 2F–I). No significant numerical changes in natural killer (NK) cell or neutrophil populations were observed (Fig. 2J and K). Within the bone marrow, femurs harvested on D19 revealed significant reduction of CD4 T-cell accumulation in either the IL2 monotherapy or the ECB combo therapy groups but only at the early D19 time point rather than when mice were sacrificed at endpoint (Fig. 2L–Q). For these experiments, it is important to note that on D19 all mice in the experiment were euthanized—regardless of current tumor burden, while, definitionally, the graphs displaying animals euthanized at endpoint does not include the mice treated with ECB which had resisted GL261 tumor growth and became long-term survivors. Splenic weight but not thymic weight was increased relative to PBS-treated controls both at D19 early in the process of tumor clearance as well as in animals euthanized when they met endpoint criteria (Fig. 2R–Y).

The majority of ECB-treated animals which lose their tumor bioluminescent signature have already done so by D21 (Fig. 1F). On the basis of this observation, we conducted flow cytometric analysis on perfused brain tissue harvested from treated GL261-Luc–bearing animals at the D19 time point, when active tumor clearance could still be occurring. Both ECB treatment and anti–PD-1 monotherapy increased the total number of CD45+ immune cells and microglia (Fig. 3A, B, and C), as well as MHC Class II expression on microglia (Fig. 3D), uncommon in homeostatic CNS conditions (35). These two treatment modalities also increased the infiltration of MHC class II positive monocytes into the CNS relative to naïve animals (Fig. 3E and F). However, only ECB therapy increased the total numbers of both CD8 and CD4 T cells within the CNS relative to untreated glioma bearing mice (Fig. 3G and H). Because the promotion of T_reg proliferation is a frequently raised concern for IL2 containing therapeutic interventions, we confirmed that while the total number of T_regs expands in the CNS compared with the counts found in untreated glioma bearing mice (Fig. 3I), the relative proportion to total CD4 T cells within the CNS is unchanged between groups (Fig. 3J). This shows ECB combination therapy to not selectively expand only the T_reg subpopulation of CD4 T cells. We next designed a flow cytometry panel to dissect cellular expression of the IL2R subunits and PD-1 expression on tumor-infiltrating T cells following ECB therapy or either monotherapy. This was intended to determine if PD-1 receptor expression was modulated by MSA-IL2 therapy and, conversely, whether IL2 receptor subunit expression was modulated by αPD-1 therapy. Such a finding would provide evidence of such synergy between these two therapeutic reagents. We observed no measurable difference in relative expression of IL2 receptor subunits or PD-1 receptor expression on CD8 (Fig. 3K–M) or CD4 T cells (Fig. 3N–P) in any of the therapy groups when brain infiltrating immune cells were analyzed at the day 19 time point, beyond a relative decrease in PD-1 positivity among both CD4 T cells in ECB-treated animals. Therefore, the reciprocal modulation of receptor expression, on T cells at least, appears not to be the primary source of synergy for this combination therapy.

Although understanding of anti–PD-1 checkpoint inhibition's mechanism is constantly evolving, a commonly expressed viewpoint emphasizes a CD8 T cell–dependent mechanism for the therapy's antitumor activity (36). This, coupled with the increased CD8 T-cell abundance in ECB Responder brains (Fig. 3G), drove us to investigate the involvement of CD8 T cells in ECB therapy. We tested this by manipulating the ability of CD8 T cells to be primed against particular MHC class I molecules. CD8 T-cell responses in WT C57BL/6 mice can be directed against both H-2D^b restricted and H-2K^b restricted antigens. Our previously published H-2D^b−/− H-2K^{b LoxP/LoxP} mice (K^b LoxP Cre-) have normal numbers of peripheral CD8 T cells but cannot generate CD8 T-cell responses against H-2D^b-restricted antigens (24). Crossing with an H-2D^b−/− H-2K^b−/− mouse expressing cytomegalovirus (CMV) promoter-driven Cre-recombinase causes ubiquitous excision of the floxed H-2K^b molecule, thus generating an entirely MHC class I–deficient mouse incapable of CD8 T-cell development (ref. 24; CMV-Cre K^b LoxP). Circulating levels of CD8 T cells were confirmed by flow cytometry in experimental animals at D14 post GL261-Luc implant, immediately before administration of ECB therapy (Fig. 4A). Unexpectedly, long-term survival following ECB treatment was not governed by host MHC class I–restricted antigen presentation or the presence of a functional CD8 T-cell response (Fig. 4B). These data show that therapeutic efficacy of ECB is independent of antitumor CD8 T cell recognition of MHC class I–restricted tumor antigens presented by host cells.

We next assessed the contribution of CD4 T cells to effective ECB therapy. CD4 T cells have been shown to contribute to antitumor activities in human GBM, with their loss being an integral feature of GBM-associated immunosuppression (37–39). In some contexts, increased CD4 T-cell activity has correlated with responsiveness to anti–PD-1 ICB (40–42). We also observed that on D19, perfused brain tissues of ECB-treated mice had an enrichment of CD4 T cells relative to PBS-treated glioma bearing mice (Fig. 3G). We therefore tested the relevance of this population to tumor clearance by depleting CD4 T cells from GL261-Luc–bearing animals on D10 post inoculation to minimally alter early tumor growth. We also depleted CD8 T cells in a separate group to further build upon our previous finding that ECB was independent of host MHC class I (Fig. 4C). In the absence of CD4 T cells, the survival advantage conferred by ECB therapy over PBS treatment was lost (Fig. 4D). The absence of CD8 T cells did not change the significant increase in survival of ECB treatment over control (Fig. 4D). We performed serial bleeds to track the efficacy of peripheral cellular depletion, with a D21 time point provided below (Fig. 4E–I). Interestingly, while the abundance of B cells in peripheral blood following ECB therapy was unaltered by CD4 or CD8 cellular depletion (Fig. 4H), the MFI of MHC Class II expression was not significantly different between ECB and Control treated animals only in the CD4 T-cell depletion group (Fig. 4I). These data combined demonstrate an unexpected critical requirement for CD4 T cells for therapeutic benefit provided by ECB therapy.

Finally, we tested our ECB therapy against the CT2A orthotopic glioma model, which is characterized as being unresponsive to αPD-1 monotherapy (43, 44). We used the same inoculation and treatment time-course as used in the GL261 model (Fig. 5A). Due to the lack of luciferase expression in CT2A glioma, we assigned mice to groups based on tumor size as measured by T2-weighted MRI at day 12 post inoculation such that tumor burden was identical between groups (Fig. 5B and C). MRI images taken longitudinally allowed for tracking of individual tumor volumes over time (Fig. 5D–G). Similar to what was observed in the GL261 glioma model, half of ECB-treated animals showed responsiveness to ECB therapy (Fig. 5D and H). Quantification of the tumor volume reduction between day 16 to day 20 clearly shows a bimodal split among the ECB-treated group between responsive and unresponsive animals (Fig. 5H). The response rate among ECB-treated animals was significant compared with controls (Fisher exact test; P = 0.0294; Fig. 5I). ECB combination therapy also generated a trending increase in survival for ECB-treated animals relative to Control and MSA-IL2 monotherapy treated animals (Fig. 5J). The raw images and tumor renders for ECB-treated long-term survivors which demonstrates their reduction in tumor volume over time can be found in Supplementary Fig. S1A.

We also interrogated immune activation in the peripheral blood of these animals on day 22 post inoculation, one day after our second dose of therapy. We observed an increase in absolute counts of monocytes in both MSA-IL2 receiving conditions relative to the PD-1 and Control groups (Fig. 5K and L). Increases in CD4 T-cell numbers in the ECB-treated group only reached significance against the PD-1 arm (Fig. 5M). Meanwhile, reductions in CD8 T cells and B cells relative to the Control arm (Fig. 5N and O) were a surprising finding not observed in the D26 blood sample in the GL261 model (Fig. 2D and F). An increase in neutrophils in MSA-IL2 receiving conditions relative to the Control arm (Fig. 5P) also did not reach significance in the GL261 model bleed (Fig. 2K), but the increase in B Cell MHC Class II in the ECB group still occurred, as before, only in that combination therapy arm (Fig. 5Q and R).

In this study, we demonstrate that the efficacy of anti–PD-1 ICB against the GL261-Luc murine glioma model can be improved through incorporating an extended half-life IL2 molecule. This finding is important against the backdrop of current therapies and challenges in GBM treatment. The highly invasive nature of GBM prevents even radical resection from being curative, and addition of radiation and systemic chemotherapy have also failed to provide cures (2, 45). Harnessing the adaptive immune system to target malignant cells has clear appeal, but has yet to achieve great efficacy in GBM tumors relative to other tumor types (13). Driving responses against GBM tumor antigens is complicated by tumor heterogeneity and the scarcity of public antigens (10, 11). Meanwhile, antigen-nonspecific immunomodulation through ICB has proved insufficient for mediating tumor control (13, 14). Among other difficulties, GBM is seldom detected early, making the experimental practice of evaluating therapies only at late time points, such as D14 and D21 in the GL261 model, critical for translational significance (31). Importantly for this study, the glioma model GL261 is only mildly immunogenic and also resists anti–PD-1 ICB (Fig. 1F and H; refs. 15, 16). Enhancing ICB with an engineered IL2 molecule boosts antitumor immunity and was curative in about 50% of treated GL261-bearing mice (Fig. 1C and F), without driving weight loss or other notable toxicities (Fig. 1D and E). The potential translational impact of this therapeutic approach is further enhanced by promising results generated in the CT2A model (Fig. 5L and M). CT2A is widely accepted as a more challenging glioma model to treat with CBI approaches like αPD-1 (17, 43, 44). Therefore, the ability to reduce tumor volume in some animals (Fig. 5D; Supplement 1) and to generate long-term survivor mice is notable (Fig. 5J). Importantly, these experiments in GL261 and CT2A harboring mice demonstrate that susceptibility to αPD-1 CBI can be enhanced by addition of an extended half-life IL2 molecule.

To narrow our focus of possible mechanisms, we measured various immunologic correlates of ECB therapy responsiveness. Our observations of immune cell proliferation in ECB-treated animals, with some ECB-treated spleens four times the weight of control organs (Fig. 2T and U), fits clinical observations of patients receiving IL2 monotherapy (46). Increased CD8 T cells within the brains of ECB Responders was intriguing (Fig. 3G), but the responsiveness of MHC class I–deficient animals to ECB therapy belies this correlation (Fig. 4B). The CD8 T cell–independent mechanism of action of ECB is surprising, helping to refine our understanding of possible mechanisms for ICB strategies. Few other studies currently suggest non-CD8 T-cell mechanisms for anti–PD-1 ICB (47, 48) or for high-dose IL2 monotherapy, with most of the latter simply adding in NK cells as additional important actors (18, 49). Our data fit the growing literature citing alternative mechanisms for the efficacy of anti–PD-1 therapy and broaden that to a combination immunotherapy setting (47, 48). A differential mechanism for anti–PD-1 ICB in a combination therapy with MSA-IL2 is not entirely unexpected, as it has been previously documented that anti–PD-1 combination with anti–CTLA-4 also adopts a different mechanism of action than either monotherapy alone (50). Turning the theoretical focus of ICB away from a CD8 T-cell heuristic will allow us to construct a clearer understanding of underlying mechanisms of immunotherapies already being used clinically.

Our data demonstrate CD4 T cells are required for ECB therapeutic effectiveness in the GL261 glioma model. In human GBM, therapy-associated depletion of CD4 T cells by chemotherapy and radiotherapy has been correlated with reduced survival, suggesting an important role for these cells in baseline GBM immune surveillance (51). This is in accordance with a recent finding in the 005 glioma model showing a mechanistic link between CD4 T cell secreted IFNγ and glioma cell susceptibility to phagocytosis by microglia (52). In the context of GL261 gliomas, a mechanistic relationship between CD4 T cells and ECB efficacy was established in our depletion experiment, although whether this functions through a similar CD4 T cell–microglia partnership remains to be shown (Fig. 4D). The expansion of peripheral myeloid cells driven by the ECB combination therapy may also provide important MHC Class II–restricted antigen presentation to CD4 T cells and play a role in these lymphocytes’ antitumor effect. An additional study in which a CD8 T cell–independent response to a checkpoint blockade approach in a glioma model also suggests the involvement of myeloid cells, especially those that have taken up resident macrophage identities within the CNS (53). Our data could support such a possibility, given the known expression of the PD-L1 molecule by glioma-associated macrophages (54) and the known impact of PD-1 blockade on myeloid skewing towards improved antigen presentation capacity (48). Future investigations will dissect the involvement of particular CD4 T-cell subsets and which auxiliary immune cells and functions they may be working to support, such as polarizing macrophages away from procancerous states (55) or activating B cells to produce antitumor antibodies (56).

ECB is effective in both the GL261 and the immunotherapy resistant CT2A glioma model, resulting in approximately a 50% response rate. Importantly ECB can be translated to clinical use. Multiple different extended half-life IL2 molecules are entering the clinic in various trials; for instance, NKTR-214 is an inactive PEGylated IL2 molecule that becomes active as its PEG moieties are shed in circulation (23). This agent is in multiple clinical trials in combination with anti–PD-1 ICB (nivolumab), including three ongoing phase III trials. The combination of NKTR-214 and nivolumab has not yet been employed to treat cancers of the CNS. The prior use of this combination immunotherapy in the clinic and the encouraging results we put forward in this manuscript make NKTR-214/nivolumab combination immunotherapy an intriguing approach for the treatment of GBM. Other engineered IL2 agents, such as THOR-707 (Sanofi), ALKS 4230 (Alkermes), or MDNA19 (Medicenna) hold similar promise (57). The potential to provide durable immunity against gliomas without necessitating further knowledge of GBM Class I restricted tumor antigens may broaden the patient pool that can benefit from immunotherapy against this highly fatal disease.

Z.P. Tritz reports grants from NIH, Koch Institute; and grants from NIH during the conduct of the study; grants from GI Innovations outside the submitted work. D.M. Wolf reports grants from NIH; and grants from Mayo Clinic-Koch Institute Cancer Solutions Team Grant during the conduct of the study. C.S. Malo reports grants from NIH (R01 NS103212); and grants from Mayo Clinic-Koch Institute Cancer Solutions Team Grant during the conduct of the study. B.T. Himes reports grants from NIH; and grants from Mayo Clinic-Koch Institute Cancer Solutions Team Grant during the conduct of the study. E.N. Goddery reports grants from NIH; and grants from Mayo Clinic-Koch Institute Cancer Solutions Team Grant during the conduct of the study. L.T. Yokanovich reports grants from NIH (R01 NS103212), Mayo Clinic-Koch Institute Cancer Solutions Team Grant. F. Jin reports grants from NIH (R01 NS103212); and grants from Mayo Clinic-Koch Institute Cancer Solutions Team Grant during the conduct of the study. I.F. Parney reports grants from Merck Pharmaceuticals outside the submitted work. D.J. Irvine reports personal fees and other support from Elicio Therapeutics, Strand Therapeutics, other support from Ankyra Therapeutics; and personal fees from Gensaic Bio outside the submitted work. No disclosures were reported by the other authors.

Z.P. Tritz: Conceptualization, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. K. Ayasoufi: Conceptualization, formal analysis, supervision, validation, investigation, visualization, writing–review and editing. D.M. Wolf: Investigation, visualization, writing–review and editing. C.A. Owens: Software, formal analysis, methodology, writing–review and editing. C.S. Malo: Investigation, writing–review and editing. B.T. Himes: Investigation, writing–review and editing. C.E. Fain: Investigation, writing–review and editing. E.N. Goddery: Investigation, writing–review and editing. L.T. Yokanovich: Investigation, writing–review and editing. F. Jin: Investigation, writing–review and editing. M.J. Hansen: Resources, writing–review and editing. I.F. Parney: Conceptualization, supervision, writing–review and editing. C. Wang: Resources, writing–review and editing. K.D. Moynihan: Resources, writing–review and editing. D.J. Irvine: Conceptualization, resources, supervision, writing–review and editing. K.D. Wittrup: Conceptualization, resources, writing–review and editing. R.M. Diaz Marcano: Resources, supervision, writing–review and editing. R.G. Vile: Conceptualization, resources, supervision, writing–review and editing. A.J. Johnson: Conceptualization, resources, supervision, funding acquisition, writing–review and editing.

We would like to thank Roman H. Khadka, a student with the Mayo Graduate School of Biomedical Sciences, for input on experimental design. Organ images in Figs. 2–5 were created with BioRender.com. This work was supported by the NIH (R01 NS103212), the Mayo Clinic-Koch Institute Cancer Solutions Team Grant. Z.P. Tritz had been supported by T32 NIH Training Grant in Basic Immunology (T32 AI07425–25) during this work.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.