Fc-enhanced anti-CTLA-4, anti-PD-1, doxorubicin, and ultrasound-mediated blood–brain barrier opening: A novel combinatorial immunotherapy regimen for gliomas

Abstract

Background

Glioblastoma is a highly aggressive brain cancer that is resistant to conventional immunotherapy strategies. Botensilimab, an Fc-enhanced anti-CTLA-4 antibody (FcE-aCTLA-4), has shown durable activity in “cold” and immunotherapy-refractory cancers.

Methods

We evaluated the efficacy and immune microenvironment phenotype of a mouse analogue of FcE-aCTLA-4 in treatment-refractory preclinical models of glioblastoma, both as a monotherapy and in combination with doxorubicin delivered via low-intensity pulsed ultrasound and microbubbles (LIPU/MB). Additionally, we studied 4 glioblastoma patients treated with doxorubicin, anti-PD-1 with concomitant LIPU/MB to investigate the novel effect of doxorubicin modulating FcγR expressions in tumor-associated macrophages/microglia (TAMs).

Results

FcE-aCTLA-4 demonstrated high-affinity binding to FcγRIV, the mouse ortholog of human FcγRIIIA, which was highly expressed in TAMs in human glioblastoma, most robustly at diagnosis. Notably, FcE-aCTLA-4-mediated selective depletion of intratumoral regulatory T cells (Tregs) via TAM-mediated phagocytosis, while sparing peripheral Tregs. Doxorubicin, a chemotherapeutic drug with immunomodulatory functions, was found to upregulate FcγRIIIA on TAMs in glioblastoma patients who received doxorubicin and anti-PD-1 with concomitant LIPU/MB. In murine models of immunotherapy-resistant gliomas, a combinatorial regimen of FcE-aCTLA-4, anti-PD-1, and doxorubicin with LIPU/MB, achieved a 90% cure rate, that was associated robust infiltration of activated CD8⁺ T cells and establishment of immunological memory as evidenced by rejection upon tumor rechallenge.

Conclusions

Our findings demonstrate that FcE-aCTLA-4 promotes robust immunomodulatory and anti-tumor effects in murine gliomas and is significantly enhanced when combined with anti-PD-1, doxorubicin, and LIPU/MB. We are currently investigating this combinatory strategy in a clinical trial (clinicaltrials.gov NCT05864534).

Key Points.

• We describe a novel immunotherapy agent, Fc-enhanced anti-CTLA-4 (FcE-aCTLA-4) agent, and its first investigation in gliomas.

• In human glioblastoma samples, FcγRIIIA was upregulated after treatment of doxorubicin with concomitant ultrasound-based blood-brain barrier opening.

• Our regimen combining anti-PD-1, FcE-aCTLA-4, and doxorubicin via LIPU/MB achieved over 90% cure rates in immunotherapy-resistant murine glioma.

• The results are the basis for a current first-in-human clinical trial (NCT05864534).

Importance of the Study.

Clinical trials for immune checkpoint blockade (ICB) in glioblastoma have shown limited efficacy due to an immunosuppressive microenvironment and poor antibody penetration of the blood-brain barrier. The first FDA-approved ICB, ipilimumab, is limited by germline polymorphisms that reduce antibody binding. To overcome this limitation, botensilimab, an Fc-enhanced anti-CTLA-4 (FcE-aCTLA-4) has been tested in clinical trials and showed unprecedented efficacy against immune-suppressive tumors, regardless of the polymorphisms. Low-dose doxorubicin with immune-modulating functions has been shown to potentiate the response to anti-PD-1. Altogether, our study highlights a novel therapeutic approach for the treatment of glioma using an FcE-aCTLA-4 that promotes durable anti-tumor responses. Notably, the therapeutic efficacy of FcE-aCTLA-4 is significantly enhanced when combined with doxorubicin, anti-PD-1, and LIPU/MB. Collectively, these results and the promising efficacy of botensilimab (FcE-aCTLA-4) observed in clinical trials provide the basis and rationale for investigating this novel immunotherapy strategy in a clinical trial where this is being conducted in the adjuvant setting in GBM patients (NCT05864534).

Glioblastoma (GBM) is the most common and most aggressive primary brain tumor in adults, with a poor prognosis and a 5-year survival averaging less than 10%.^1–3 Despite preclinical evidence suggesting the promise of immunotherapy, its clinical application has yielded disappointing results in unselected GBM patients.^4–6 Several factors may contribute to this discrepancy, including the blood-brain barrier (BBB) that limits effective drug delivery to regions of infiltrative disease within the peritumoral brain,^7–9 the immunosuppressive tumor microenvironment that is marked by inherently low tumor immunogenicity and a paucity of lymphocyte infiltration,^10–13 and the inter- and intratumor heterogeneity that results in an inconsistent and unpredictable response to therapies including immunotherapy.^14–18

Immune checkpoint blockade (ICB) targeting cytotoxic T-lymphocyte antigen 4 (CTLA-4) is an established form of immunotherapy and has been investigated in clinical trials for GBM (NCT02311920, NCT04396860).^19,20 Anti-CTLA-4 (aCTLA-4) bound to Tregs expressing this receptor, are designed to co-engage, and activate fragment crystallizable gamma receptors (FcγRs) expressed on macrophages or nature killer (NK) cells to promote antibody-dependent cellular phagocytosis (ADCP) or cytotoxicity (ADCC), respectively.^21–23 This form of ICB relies partly on T cell priming,²¹ and is also thought to promote the depletion of intratumoral immunosuppressive regulatory T cells (Tregs); however, it is not clear whether Treg depletion can be achieved by conventional aCTLA-4 in the clinical setting.^24,25

Botensilimab, a next-generation aCTLA-4 antibody was engineered with point mutations in the Fc region (S239D/A330L/I332E; DLE) to increase affinity for FcγRIIIA and enhance effector functions compared to wild-type Fc was developed with enhanced NK cell-mediated ADCC of Tregs.^21,26,27 In phase I clinical trials, botensilimab, used as monotherapy or in combination with balistilimab (aPD-1), demonstrated unprecedented activity in immune-resistant “cold” tumors such as microsatellite stable colon cancer and in other heavily pretreated patients who had not responded to prior immunotherapy.^28,29

The BBB is a major impediment to successful drug therapy in the brain. The core of GBM exhibits BBB breakdown and is visible on MRI due to gadolinium-based enhancement. This region is presumably relatively permeable to systemic drugs. Yet this portion of the tumor is commonly resected as part of the standard of care. GBM recurrence is mostly arising from infiltrating tumor cells residing in the peritumoral brain, where the BBB is intact, and penetration of systemic agents is poor. To address the limited penetration of systemically delivered ICB antibodies and chemotherapy drugs across the BBB, we investigated the use of low-intensity pulsed ultrasound in combination with microbubbles (LIPU/MB). This procedure temporarily opens the BBB leading to a significant increase in the concentration of systemically administered drugs and antibodies in the brain as shown in preclinical models as well as initial trial results.^30–36

Several groups have reported immune modulatory properties of doxorubicin (DOX), and how this anthracycline can enhance the efficacy of anti-PD-1 (aPD-1) ICB in cancer.^33,37,38 In particular, we showed that when delivered with concomitant LIPU/MB BBB opening, aPD-1 and DOX can modulate the tumor microenvironment in GBM patients.³⁶

In this study, we explore the use of a mouse surrogate of botensilimab, referred to as FcE-aCTLA-4, in preclinical immune-resistant GBM models, and investigated the expression of FcγRIIIA on TAMs following DOX exposure in a cohort of GBM patients treated with ICB with DOX, using LIPU/MB to enhance the delivery of these agents. Our results provide insights into novel anti-CTLA-4 antibodies engineered to harness innate immune cells like microglia for improved GBM immunotherapy. These results are the basis for a clinical trial in newly diagnosed GBM where we are investigating the therapeutic efficacy of this approach (NCT05864534).

Materials and Methods

Cell Culture

Mouse glioma cell line GL261 was obtained from the NIH, CT-2A cell line was purchased from Millipore. Both cell lines were cultured in DMEM (Gibco) supplemented with 10% FBS (Gibco) and 1% penicillin/streptomycin (Gibco) in 5% CO₂ incubators at 37°C. QPP4 cells derived from Nestin-CreER^T2 Qk^L/L; Trp53^L/L; Pten^L/L mice (QPP) were kindly provided by Dr. Amy B. Heimberger (Northwestern University).^39,40 QPP4 cells were maintained in DMEM/F-12 (Gibco) supplemented with B-27 (Gibco), 20 ng/mL rEGF (Peprotech), and 20 ng/mL rbFGF (Peprotech). Human microglial cell line HMC3 was obtained from ATCC and maintained in EMEM supplemented with 10% FBS and 1% penicillin/streptomycin in 5% CO₂ incubators at 37°C.

Intracranial Orthotopic Murine Glioma Cell Injection

All mouse protocols involved in this study were approved by Northwestern University IACUC under approval number IS00017189. Six- to eight-week-old wild-type C57BL/6 mice were purchased from Charles River Laboratories. Prior to use, all murine cell lines were tested for mycoplasma and confirmed negative. For intracranial injection, mice were anesthetized with a ketamine/xylazine cocktail. Using a stereotaxic device (Harvard Apparatus), a burr hole was drilled 3 mm lateral and 2 mm caudal to the bregma. 100 000 cells of GL261 or QPP4, and 75,000 cells of CT-2A were injected into the left hemisphere at a depth of 3 mm. Mice were monitored at least twice a week and were euthanized when they approached the endpoint based on the criteria (weight loss is >20% of pretreatment body weight or loss of mobility or severe neurological disabilities such as seizure, circular motion, etc.) following the guidelines by IACUC protocol.

Antibodies and Mouse Fc-Fcγr Cellular Binding

Fc-engineered mouse reactive anti-CTLA-4 antibodies were provided by Agenus. The sequences of the variable regions of the heavy and light chains of anti-mouse CTLA-4, clone 9D9 (RRID: AB_10949609) were used to generate surrogate antibodies with the constant regions of mouse IgG2b (mIgG2b) and mutated mIgG2b with amino acid substitutions S239D/A330L/I332E (mIgG2b.DLE; Fc-enhanced). Antibodies were produced using recombinant DNA technology in a Chinese Hamster Ovary (CHO) mammalian cell expression system. InVivoPlus-grade isotype control (clone MPC-11, mIgG2b) and mouse reactive anti-PD-1 (RMP1-14; Rat IgG2a, κ) were obtained from BioXcell (USA). CHO cells genetically engineered to express mouse FcγRI, FcγRIIB, FcγRIII, and FcγRIV were obtained from Collection de Cultures de Microorganisms, Institut Pasteur (CNCM).⁴¹ The mouse FcγR-expressing CHO cells were incubated with titrated concentrations of anti-CTLA-4 antibodies or an isotype-negative control antibody (mIgG2b). Following a 1-hour incubation, binding of the antibodies was detected by LSRFortessa flow cytometry (BD).

SPR Analysis

The affinity of anti-mouse CTLA-4 antibodies to mouse FcγRIIB and FcγRIV proteins, and anti-human CTLA-4 antibodies to human FcγRIIB, FcγRIIIA F158, and FcγRIIIA V158 proteins were determined by SPR using a Biacore T200 instrument (Cytiva) and a CM5 Biacore sensor chip immobilized with an anti-His antibody (Invitrogen). Binding kinetic analyses were carried out using Biacore evaluation software (GE Healthcare version 3.0).

Immunophenotype Analysis

Splenocytes or Percoll (Cytiva) gradient-enriched cells from the brain were filtered with a 70 µm cell strainer and incubated with mouse Fc block. Cells then underwent surface staining with primary antibodies and live/dead. After fixation and permeabilization (Foxp3/Transcription Factor Staining Buffer Set, eBioscience), intracellular staining was performed. The reagents are listed in Supplementary Table 1. Flow cytometry data were acquired by the BD Symphony and analyzed by FlowJo 10.8.1 (BD).

In Vitro TAM Phagocytosis Assay

Splenocytes were prepared in single-cell suspensions from C57BL/6-Foxp3-GFP mice. CD4⁺ T cells were sorted by CD4-PE and Foxp3-GFP expression using a BD FACS Aria II cell sorter (BD), and the cells were activated and expanded with 2000 U/mL recombinant IL-2 (Peprotech) and Gibco Dynabeads Mouse T-Activator CD3/CD28 (Invitrogen) for 3 days. TAMs were generated as described previously.⁴² Briefly, bone marrow progenitor cells were obtained from mice’s tibias and femurs by removing the epiphyses of the bones and perfusing. After lysis of RBCs using ACK buffer (Gibco), cells were placed in RPMI-1640 (Corning) with l-glutamine, 10% FBS (Gibco), 1% penicillin/streptomycin (Gibco), 1% HEPES (Gibco), 1% NEAA (Gibco), 1% sodium pyruvate (Gibco), 0.1% 2-mercaptoethanol (Gibco), and 40 µg/mL rGM-CSF. In vitro phagocytosis assay was accomplished using the IncuCyte System. CellTracker Red CMTPX (Thermo, 1:2000) stained 2.5 × 10⁵ TAMs were plated in a 24-well plate and incubated overnight for adherence. GFP-labeled Tregs and non-Treg CD4⁺ T cells (1.25 × 10⁵ cells) were added with isotype antibody or FcE-aCTLA-4 antibody. The cells were incubated in complete RPMI-1640 for 18 hours. Phagocytotic cells are displayed as red/green overlapped object counts per image every hour.

Collection of Paired GBM Patients’ Biopsy Samples

All the biopsy samples presented in this study are from patients enrolled in our phase I/II clinical trial (NCT04528680).³⁴ During surgery for tumor resection, a skull-implantable ultrasound device (SonoCloud-9, CarThera) was placed at the craniectomy site and subsequently activated during cycles of chemotherapy for opening of the BBB and delivery of albumin-paclitaxel. Systemic injections of 10 µL/kg of microbubbles (DEFINITY, Lantheus) were administered immediately before activation of the Sonocloud-9. As part of single-patient expanded access protocols conducted in accordance with the institutional ethical regulations and the declaration of Helsinki principles, approved by the FDA and by the Institutional Review Board (IRB) at Northwestern University, a subset of patients that progressed under this therapy received pembrolizumab (anti-PD-1, 200 mg) and doxorubicin (30 mg) with LIPU/MB. Patients then received surgery for resection or biopsy of the recurrent tumor within 48 hours of therapy. For these patients, we had biopsies collected before therapy with DOX and pembrolizumab, and during therapy with these 2 drugs. Detailed information is depicted in Supplementary Table 2. GBM samples obtained after neoadjuvant administration of anti-PD-1 blockade, were obtained at the University of California San Francisco, under an IRB protocol approval.

Monocyte Isolation From Human PBMC and M1, M2 Differentiation

M1 macrophages are generated by isolating monocytes from PBMCs and maintaining them in IMDM + 10% Human AB serum + 50 ng/mL M-CSF for 7 days. The macrophages are then polarized to M1-like phenotype by incubating for 48 hours in IMDM + 10% Human AB serum + 50 ng/mL GM-CSF, 50 ng/mL IFN-γ, and 100 ng/mL LPS. M1 macrophages and HMC3 microglia cells are collected, washed in macrophage assay buffer (MAB: IMIM + 10% AB serum) for macrophages or assay buffer (AB: EMEM + 10% FBS) for HMC3 cells. Cells are then plated in a 96-well plate and Doxorubicin (Selleckchem) is added as a 1:2 serial dilution (30–0.23 μM) and incubated at 37°C.

Multiplex Immunofluorescence Staining/Analysis of Multispectral Images

Sections of 5-µm thickness were obtained from FFPE-embedded tumor tissues. Deparaffinization of the slides was achieved using BOND dewax solution followed by heat-induced epitope retrieval using BOND epitope retrieval solution (pH6) or pH9 EDTA buffer for 20 minutes. 3,3’-diaminobenzidine chromogen staining was initially performed to determine the optimal concentrations of each antibody in human GBM tissues. The list of primary antibodies and dilution factor was listed in Supplementary Table 3. Multiplex staining was performed in multiple cycles involving a heat-induced epitope retrieval step, protein blocking, epitope labeling, and signal amplification. Once all markers were stained, spectral DAPI was used to counterstain the slides, which were mounted using Prolong Diamond Antifade Mountant. Multispectral imaging (MSI) was performed using the Vectra 3 Automated Quantitative Pathology Imaging System from Akoya Biosciences. First, whole slide images were acquired after autoadjusting focus and signal intensity. Then, MSI was acquired in the tumor regions delineated by a certified neuropathologist at 20× of the original magnification. For analysis of MSI, we created a spectral library for all Opal dyes to subject acquired multispectral images to spectral unmixing that enabled the identification and separation of weakly expressing and overlapping signals from background to visualize the signal of each marker in inForm Tissue Finder software (inForm 2.6, Akoya Biosciences). Using InForm, the adaptive cell segmentation feature was used to identify the nucleus of the analyzed cells and to determine the nuclear and cytoplasmic compartments on each cell. A machine-learning algorithm within inForm was used in which cells were automatically assigned to a specific phenotype. Batch analysis was used to analyze all tumor samples under the same segmentation and phenotype settings. The processing and analysis of images from all tumor samples were exported to cell segmentation tables. Exported files from inForm were processed in R using R packages Phenoptr and PhenoptrReports to merge and create consolidated single files for each tumor sample. Consolidated files had cell phenotypes as outputs that we employed for further quantification and spatial analyses using the Phenoptr R addin.

Low-Intensity Pulsed Ultrasound and Microbubbles

The LIPU/MB sonication procedure for mouse experiments was performed using a preclinical LIPU device (SonoCloud Technology) manufactured by CarThera and was previously described.³⁰ Mice were anesthetized with a ketamine/xylazine cocktail. Microbubbles (MB, Lumason, Bracco) were reconstituted according to the manufacturer’s instructions and injected at a dose of 7.5 mL/kg through the retro-orbital route. Shortly after MB administration, mice were placed supine upon the ultrasound transducer holder, and the sonication began. A 1 MHz, 10-mm-diameter flat ultrasound transducer was fixed in a holder filled with degassed water, and sonication was performed transcranially. Sonication was performed for 60 seconds using a 25 000-cycle burst at a 1 Hz pulse repetition frequency and an acoustic pressure of 0.3 MPa as measured in water.

Statistical Analyses

Statistical analyses were performed using Prism Software 9.4.1 (GraphPad). Unpaired Student’s t-test was used to compare statistical differences between the 2 groups. For Kaplan–Meier survival curves, the log-rank (Mantel–Cox) test was adapted to determine the significance between groups. Statistical significances were presented in P value, or P < .05 was considered significant, *P < .05, **P < .01, ***P < .001, ****P < .0001.

Results

FcγR Binding Characteristics of an Fc-Enhanced Anti-Mouse CTLA-4 Antibody Surrogate

Firstly, we compared the FcγR binding characteristics of a human IgG1 FcE-aCTLA-4 (hIgG1.DLE) and an IgG2b mouse aCTLA-4 antibody engineered with the same mutations in the Fc region as botensilimab, to an unmodified human IgG1 and mouse IgG2b variant, respectively (Figure 1, Table 1). The binding to activating (A) and inhibitory (I) IgG Fc receptors by surface plasmon resonance (SPR) was used to establish the A/I ratio (Tables 2 and 3), a measure shown to be predictive of in vivo cytotoxicity.⁴³

Figure 1.

Fc-enhanced anti-mouse CTLA-4 antibody exhibits increased binding to Fcγ receptors. Binding kinetics of an unmodified (mIgG2b) anti-mouse CTLA-4 (aCTLA-4) and an Fc-enhanced (mIgG2b-DLE) anti-mouse CTLA-4 variant to mouse FcγRIIB and FcγRIV proteins using SPR. For binding to mouse FcγRIIB (A and B), antibodies were injected at concentrations ranging from 62 nM to 8 µM. For binding to mouse FcγRIV (C and D), antibodies were injected at concentrations ranging from 0.93 to 120 nM. Assessment of mouse reactive anti-CTLA-4 antibody binding to Chinese hamster ovarian (CHO) cells genetically engineered to express mouse (E) Fcγ receptor (FcγR) I, (F) FcγRIIB, (G) FcγRIII, or (H) FcγRIV. Cells were incubated with increasing concentrations of Fc-enhanced mouse reactive anti-CTLA-4 antibody (clone 9D9, mIgG2b.DLE), anti-CTLA-4 mIgG2b, or an mIgG1 isotype control antibody (negative control). Binding was analyzed by flow cytometry using a fluorochrome-conjugated anti-mouse F(ab’)₂ secondary antibody.

Table 1.

Summary of Fc Isotype, Fc Mutations and Fcγ Receptor (FcγR) Binding Characteristics to Human FcγRIIIA or Mouse FcγRIV of Anti-Human and Anti-Mouse CTLA-4 Antibodies

	Anti-CTLA-4 antibody	Fc isotype	Fc mutations	FcγR binding characteristics
Human	Parental	IgG1	—	Low
	Botensilimab (Fc-enhanced)	IgG1.DLE	S239D.A330L.I332E	>FcγRIIIA binding
Murine surrogates	9D9 Parental	mIgG2b	—	Low
	9D9 FcE (Fc-enhanced)	mIgG2b.DLE	S241D.A332L.I334E	>FcγRIV binding

Table 2.

Binding Affinities of Anti-Human CTLA-4 Antibodies to Human FcγRIIB, FcγRIIIA F158, and FcγRIIIA V158 Proteins Determined by SPR. The Ratio of Binding of an IgG Subtype to Activating FcγRs and Inhibitory FcγRIIB Known as the Activating to Inhibitory (A/I) Ratio Are Shown

Human	FcγRIIB	FcγRIIIA F158	FcγRIIIA V158	IIIA F158/IIB	IIIA V158/IIB
Antibody	KD (M)	KD (M)	KD (M)	A/I	A/I
hIgG1	6.86E-06	9.19E-08	2.76E-08	75	249
hIgG1-DLE	3.23E-06	2.83E-09	2.76E-09	1141	1170

Table 3.

Binding Affinities of Anti-Mouse CTLA-4 Antibodies to Mouse FcγRII and FcγRIV Proteins Determined by SPR. The Ratio of Binding of an IgG Subtype to Activating FcγRs and Inhibitory FcγRIIB Known as the Activating to Inhibitory (A/I) Ratio Are Shown

Mouse	FcγRIIB	FcγRIV	IV/IIB
Antibody	KD (M)	KD (M)	A/I
mIgG2b	5.80E-06	6.13E-08	95
mIgG2b-DLE	7.11E-06	1.70E-09	4182

We observed that the mouse FcE-aCTLA-4 bound with higher affinity to FcγRIV, the mouse ortholog of human FcγRIIIA, with an equilibrium dissociation constant (K_D) of 1.70 nM, compared to 61.3 nM for the unmodified mouse IgG2b aCTLA-4 antibody (Figure 1A and B). However, binding affinity to the inhibitory mouse FcγRIIB was similar between FcE-aCTLA-4 (K_D = 7.11 μM; Figure 1C) and the unmodified aCTLA-4 (K_D = 5.88 μM; Figure 1D). Improved binding by mouse FcE-aCTLA-4 to mouse FcγR, measured by SPR, was confirmed by flow cytometry using CHO cells expressing mouse FcγRs. FcE-aCTLA-4 demonstrated superior potency and maximal binding to cell-expressed mouse FcγRI, FcγRIIB, FcγRIII, and FcγRIV (Figure 1E–H) compared to the unmodified mouse IgG2b aCTLA-4 antibody.

Similarly, the human FcE-aCTLA-4 antibody bound with significantly higher affinity than the parental human IgG1 for FcγRIIIA, with a ~2-fold increase in binding to human FcγRIIB (Table 2; Supplementary Figure 1). Notably, the human FcE-aCTLA-4 antibody showed significantly improved binding to the low-affinity FcγRIIIA polymorphic variant (F158) with an average K_D of 2.83 nM compared to 91.9 nM for the corresponding unmodified human IgG1 variant (Table 2). Binding to the high-affinity FcγRIIIA polymorphic variant (V158) was also improved with the human FcE-aCTLA-4 antibody (K_D = 2.76 nM) compared to the unmodified human IgG1 aCTLA-4 antibody (K_D = 27.6 nM) (Table 2; Supplementary Figure 1).

Given that the ratios of K_D between activating FcγR (FcγRIIIA or FcγRIV) and inhibitory FcγRIIB were higher in both human and mouse FcE-aCTLA-4, suggesting enhanced FcγR-mediated effector functions and potentially improved therapeutic efficacy (Tables 2 and 3).

FcγRIIIA Expression in Newly Diagnosed Versus Recurrent Glioblastoma Microenvironment

We characterized the expression of FcγRIIIA and CTLA-4 in GBM in single-cell RNA-sequencing (scRNA-seq) dataset of human GBM (Figure 2A).⁴⁴ Unsupervised clustering was performed to identify various cell types. We then evaluated the expression of FcγRIIIA and CTLA-4 in these populations and observed that FcγRIIIA was mostly expressed by myeloid cells, including granulocytes, macrophages, microglia, and monocytes. In contrast, CTLA-4 was preferentially expressed by T cells and Tregs (Figure 2B and C). Similarly, we further investigated published mouse CT-2A glioma scRNA-seq data to explore the expression of differences. Following a similar approach as with human GBM samples, we performed unsupervised clustering to characterize various cell types isolated from mouse brains (Supplementary Figure 2A and B).⁴⁵ Consistent with the expression pattern in humans, Fcgr4 (FcγRIV), the mouse ortholog of human FCGR3A, was preferentially expressed in TAMs (Supplementary Figure 2C), whereas CTLA-4 was found to be highly expressed by T cells, particularly by Tregs (Supplementary Figure 2D).

Figure 2.

FcγRIIIA expression in newly diagnosed versus recurrent glioblastoma microenvironment. (A) Uniform manifold approximation and projection (UMAP) dimensionality reduction plot indicating cell categories from single-cell RNA-sequencing analysis of human glioma patients. Violin plot of FCGR3A (B) and CTLA-4 (C) expression across the cell types in human recurrent glioblastoma. (D) Representative multiplex immunofluorescence images of a newly diagnosed glioblastoma and recurrent glioblastoma from patients showing FcγRIIIA expression on myeloid cell compartment. (E) Quantification graph showing the percentages of infiltrating macrophage/microglial cell phenotypes out of FcγRIIIA⁺ cells. An unpaired t-test was used for statistical analysis. Quantitative cell density analysis of 3 different myeloid phenotypes with FcγRIIIA including TMEM119⁺CD163⁻ (F), CD11c⁺HLA-DR⁺ (G), and CD163⁺TMEM119⁻ (H). Quantitative cell density analysis of 3 different myeloid phenotypes, including TMEM119⁺CD163⁻ (I), CD11c⁺HLA-DR⁺ (J), and CD163⁺TMEM119⁻ (K). An unpaired t-test was used for statistical analysis. Data indicate mean ± SD and the P value is depicted.

To verify these findings at the protein level, we used multiplex immunofluorescence on specimens from newly diagnosed GBM (nGBM) and recurrent GBM (rGBM) patients (Figure 2D). For this, we first evaluated the density of TAM phenotypes in nGBM and rGBM (Supplementary Figure 3A). We then assessed the percentage of FcγRIIIA⁺ cells corresponding to each phenotype (Figure 2E). Utilizing scRNA-seq data, we identified microglia as TMEM119⁺CD163⁻ cells, antigen-presenting cells (APCs) as CD11c⁺HLA-DR⁺, and macrophages as CD163⁺TMEM119⁻ (Supplementary Figure 4). We then compared the density of FcγRIIIA⁺ cells across different TAM phenotypes between nGBM and rGBM, and observed significantly higher densities of microglia, as well as APCs, but not macrophages (Figure 2F–H). We encountered differences in the overall abundance of these phenotypes between these 2 disease stages, regardless of FcγRIIIA expression (Figure 2I–K) but observed no changes in the percentage of these phenotypes that were FcγRIIIA⁺ (Supplementary Figure 3B). These results suggest that the differences in FcγRIIIA⁺ cells encountered between nGBM and rGBM are explained by the relative loss of microglia, which expresses this receptor, and infiltration of myeloid-derived macrophages during tumor progression.

Fc-Enhanced Anti-CTLA-4-Mediated Treg Depletion in the Glioma Microenvironment

We assessed whether the enhanced affinity of mouse FcE-aCTLA-4 to FcγRIV could deplete Tregs within the glioma tumor microenvironment. First, an in vitro phagocytosis assay was performed with TAM generated by exposing BMDM to CT-2A-conditioned media. These TAMs were then co-cultured with GFP-expressing splenic CD4⁺Foxp3⁺ Tregs or CD4⁺Foxp3⁻ non-Tregs in the presence of FcE-aCTLA-4, the unmodified parental aCTLA-4, or Isotype control (IC) IgG antibodies (Figure 3A). In this assay, FcE-aCTLA-4 promoted phagocytosis of CD4⁺Foxp3⁺ Tregs compared to the IC IgG antibody, which was not observed with parental aCTLA-4 (Figure 3B). Depletion was specific to CD4⁺Foxp3⁺ Tregs, as we did not observe any phagocytosis of the CD4⁺Foxp3⁻ non-Tregs (Figure 3C), consistent with the lower cell surface expression of CTLA-4 on non-Treg cells.⁴⁶

Figure 3.

Fc-enhanced anti-CTLA-4-mediated Treg depletion in glioma microenvironment. (A) Schematic illustration of cell preparation. TAMs generated from bone marrow progenitor cells differentiated with M-CSF and conditioned with CT-2A mouse glioblastoma cell cultured media. (B) Graph showing time-dependent cell overlap (phagocytosis) between Tregs (② right in A) and macrophages (① in A) with isotype control, parental anti-CTLA-4, and FcE anti-CTLA-4 antibody. (C) Graph showing time-dependent cell overlap (phagocytosis) between CD4⁺ T cells (non-Tregs, ② left in A) and macrophages (① in A) with isotype control, parental anti-CTLA-4, and FcE anti-CTLA-4 antibody. (D) Schematic illustration of immunophenotype analysis design. (E) Flow cytometry analysis showing Tregs (Foxp3⁺) in CD4⁺ T cells at day 21. (F) Tumor-specific Treg ratio was plotted at 2 different time points (left) and the Treg ratio in the spleen on days 14 and 21 (right). Each comparison was analyzed by unpaired Student’s t-test. (G) Time-dependent PD-1 downregulation in CD4⁺ T cells. At each time point, the mean fluorescent intensity of PD-1 was calculated and plotted (left), and the representative histogram (right). (H) Time-dependent PD-1 downregulation in CD8⁺ T cells. At each time point, the mean fluorescent intensity of PD-1 was calculated and plotted (left), and the representative histogram (right). An unpaired t-test was used for statistical analysis. Data indicate mean ± SD, and significance is depicted as ns: not statistically significant, *P < .05, **P < .01, ***P < .001, ****P < .0001.

To confirm the ability of FcE-aCTLA-4 to selectively deplete intratumoral Tregs in vivo, CT-2A tumor-bearing mice were treated with FcE-aCTLA-4, parental aCTLA-4, or IC IgG antibodies and immunophenotyped the brain-resident and systemic lymphocytes on days 14 and 21 post-tumor implantations with flow cytometry (Figure 3D, gating strategy: Supplementary Figure 5). Consistent with our observations in vitro, FcE-aCTLA-4 promoted superior depletion of intratumoral Tregs compared to parental aCTLA-4 or IC IgG. This effect was more pronounced at day 21 compared to day 14 post-tumor implantation (Figure 3E and F, left). In contrast, splenic Tregs were not affected by treatment (Figure 3F, right). Notably, the FcE-aCTLA-4-mediated depletion of intratumoral Tregs was associated with a significant decrease in PD-1 expression by tumor-infiltrating CD4⁺ (Figure 3G) and CD8⁺ T cells (Figure 3H), whereas PD-1 expression was unaltered in splenic lymphocytes (Supplementary Figure 6). We did not observe upregulation of IFN-γ or granzyme B expression in CD8⁺ T cells but found an increase in Arg1 expression in TAM in the context of FcE-aCTLA-4 (Supplementary Figure 7).

Anti-Tumor Efficacy of Fc-Enhanced Anti-CTLA-4 Antibody in Murine Glioma Models

To investigate the efficacy of this ICB antibody, survival studies were conducted in 3 immune-competent murine GBM models: GL261, CT-2A, and QPP4, with the latter 2 known for immunotherapy resistance.^40,47,48 Mice were treated with FcE-aCTLA-4, parental aCTLA-4, or IC antibodies (Figure 4A). In GL261, FcE-aCTLA-4 marginally increased median survival to 21 days from 18 days compared to IC-treated mice (P = .0090; Figure 4B), while it significantly enhanced long-term survival to 70% and 80% in QPP4 (P = .0002; Figure 4C) and CT-2A (P < .0001; Figure 4D, left) models, respectively. In contrast, parental aCTLA-4 showed minimal effect.

Figure 4.

Anti-tumor efficacy of Fc-enhanced anti-CTLA-4 antibody in murine glioma. (A) Schematic illustration of treatment and survival study design. The efficacy test of the FcE anti-CTLA-4 antibody was conducted in 3 different mouse syngeneic models GL261, QPP4, and CT-2A. (B) Kaplan–Meier survival curve of GL261-bearing C57BL/6 mice treated with FcE anti-CTLA-4 or parental antibody. (C) Kaplan–Meier survival curve of QPP4 bearing C57BL/6 mice treated with FcE anti-CTLA-4 or parental antibody. (D) Kaplan–Meier survival curve of CT-2A model (left) and tumor rechallenge survival curve from long-term survivors (right). (E) Flow cytometry analysis of immune profiles for comparison of non-treat control tumor and long-term survivors. The myeloid cells and lymphocytes ratio was plotted (left), and CD8/CD4 ratio was plotted (right). (F) Representative CD8⁺ T cell properties were evaluated by PD-1 and IFN-γ expression (left) and a summarized bar graph (right). (G) CD8 immunohistochemistry images of newly developed non-treat control CT-2A tumor (left) and long-term survivor mice (right). (H) Foxp3 immunohistochemistry images of newly developed non-treat control CT-2A tumor (left) and long-term survivor mice (right) (scale bar = 100 µm). Unpaired t-test (E) or 2-way ANOVA (F) was used for statistical analysis. Data indicate mean ± SD, and significance is depicted as ns = not statistically significant, *P < .05, **P < .01, ***P < .001, ****P < .0001. LTS (long-term survivors).

To evaluate aCTLA-4 therapy’s potential to promote immune memory and prevent glioma recurrences, mice cured of CT-2A tumors were re-challenged 120 days post-initial implantation with a new contralateral intracranial glioma implant, without further treatment, to assess survival. All FcE-aCTLA-4-treated mice survived subsequent tumor re-challenge, suggesting that FcE-aCTLA-4 treatment stimulated a long-lasting immune memory response capable of rejecting the second tumor implant (Figure 4D, right). We analyzed the brains of long-term survivors who underwent a secondary tumor challenge. Using flow cytometry, we profiled brain-resident immune cells. Long-term survivors exhibited a decreased myeloid-to-lymphocyte ratio (P = .0273; Figure 4E, left) and an increased CD8⁺ to CD4⁺ T cell ratio (P < .0001; Figure 4E, right) compared to untreated controls. Additionally, CD8⁺ T cells from treated mice showed high IFN-γ production and low PD-1 expression (P = .0434, Figure 4F), indicating an activated state. Immunohistochemistry confirmed that the elevated tumor-infiltrating CD8⁺ T cells were concentrated in the gliotic or lesional brain regions (Figure 4G). Consistent with the enhanced effector function by FcE-aCTLA-4, there was also a notable reduction in Foxp3⁺ Tregs in the treated mice (Figure 4H).

To examine T cell phenotype in gliomas after FcE-aCTLA-4 treatment, we compared Ki67 expression in CD8⁺ T cells from long-term survivors and untreated CT-2A tumor-bearing brains. Following rechallenge on day 120 by injecting CT-2A cells into the contralateral hemisphere, Ki67 levels in CD8⁺ T cells were elevated 7 days later, indicating a potentiated immune memory and cell proliferation, absent in cervical lymph node CD8⁺ T cells (Supplementary Figure 8). Further, survival studies in CD8 knockout (CD8 KO) mice and mice treated with the CSF1R inhibitor PLX3397, which depletes TAMs,^49,50 showed no FcE-aCTLA-4 efficacy in these models compared to untreated controls (Supplementary Figure 9A–C). A marked depletion of Iba1 + brain macrophages/microglia was confirmed in the PLX3397-treated groups (Supplementary Figure 9G and H). This supports that the efficacy of FcE-aCTLA-4 in the mouse CT-2A glioma model relies on CD8⁺ T cells and TAMs.

Doxorubicin Plus Anti-PD-1-Mediated Alteration of FcγRIIIA-Related Phenotype of Glioblastoma Infiltrating Myeloid Cells

To investigate whether DOX enhances FcE-aCTLA-4 efficacy, we investigated its effect with LIPU/MB on FcγRIIIA expression in human GBM. We recently described a recurrent GBM patient cohort treated with DOX, aPD-1, and LIPU/MB.³⁶ Briefly, after recurrence following paclitaxel-based trial-related treatment (NCT04528680),³⁴ 4 patients received induction with liposomal DOX (30 mg) and LIPU/MB as part of a single-patient expanded access protocol and, within 10–14 days, underwent a second dose of DOX (30 mg) and aPD-1 (pembrolizumab 200 mg), with concomitant LIPU/MB. All 4 patients underwent 1–3 cycles of DOX/aPD-1 plus LIPU/MB every 3 weeks, followed by surgery for tumor resection or biopsy. This process allowed for the availability of tissue prior to treatment with DOX and aPD-1 (pre-DOX GBM samples) obtained at the time of SC9 implant (during surgery for clinical trial) and tumor tissue resected after 1–3 cycles of DOX and aPD-1 (during-DOX GBM samples). In all cases, surgery was performed 2 days after DOX administration with LIPU/MB (Figure 5A).

Figure 5.

Doxorubicin plus anti-PD-1-mediated alteration of FcγRIIIA-related phenotype of glioblastoma infiltrating myeloid cells. (A) Clinical course of recurrent GBM patients analyzed in this study. Patients underwent surgery for tumor resection (pre-DOX samples) and skull implantation of the SonoCloud-9 ultrasound device for treatment with previous chemotherapy. Upon tumoral progression, induction treatment with DOX delivered with LIPU/MB was initiated to treat the recurrent tumor, followed by additional treatment cycles with both liposomal DOX and anti-PD-1 delivered by LIPU/MB. The tumor exposed to these therapies was resected (during-DOX samples) and further analyzed. (B) Representative multiplex immunofluorescence images of tumor regions before and during-DOX treatment. (C) Quantification graph showing the percentages of infiltrating macrophage/microglial cell phenotypes out of FcγRIIIA⁺ cells before and during-DOX treatment. Comparison of infiltrating TMEM119⁺FcγRIIIA⁺ cells (D), CD163⁺FcγRIIIA⁺ cells (E), and CD11c⁺FcγRIIIA⁺ cells (F) before and during-DOX treatment. Paired t-test was used for statistical analysis and, the P value was indicated. Quantitative mean fluorescent values for FcγRIIIA were measured from multiplex immunofluorescence in TMEM119⁺ cells (G), CD163⁺ cells (H), and CD11c⁺ cells (I). An unpaired t-test was used for statistical analysis. Data indicate mean ± SD and the P value is depicted.

Utilizing 4 pre-DOX and during-DOX paired samples, we performed multiplex immunofluorescent analysis (Vectra 3, Akoya Bioscience) to assess the impact of this treatment on the glioblastoma microenvironment (Figure 5B). Overall, after DOX/aPD-1 treatment with concomitant LIPU/MB, the proportion of infiltrating myeloid cells expressing TMEM119, CD163, or CD11c expression increased and represented approximately 37% of all cells (Supplementary Figure 10). Quantitative multiplex immunofluorescence revealed that treatment increased the frequency of TMEM119⁺FcγRIIIA⁺ (P = .0254, Figure 5D), CD163⁺FcγRIIIA⁺ (P = .0549, Figure 5E), and CD11c⁺FcγRIIIA⁺ (P = .0071, Figure 5F) TAMs compared to pre-DOX samples. Additionally, myeloid cells expressing CD163 and CD11c showed elevated FcγRIIIA in during-DOX samples (Figures 5H and I), but not observed in TMEM119⁺ cells (Figure 5G). Another analysis using COMET (Lunaphore) further confirmed the effect of DOX, revealing increased FcγRIIIA expression on CD68⁺ and CD11c⁺ myeloid cells in during-DOX samples (Supplementary Figure 11). We then explored whether anti-PD-1 or LIPU/MB alone could produce similar effects on FcγRIIIA. To investigate this, we examined human recurrent GBM tumor samples treated with neoadjuvant anti-PD-1 therapy. Using multiplex immunofluorescence, we compared immunotherapy-naive tumors, tumors treated with neoadjuvant anti-PD-1, and tumors that received both anti-PD-1 and DOX. However, we did not observe any significant increase in FcγRIIIA expression in the tumors treated with neoadjuvant anti-PD-1 (Supplementary Figure 12).

Enhanced Efficacy of Fc-Enhanced Anti-CTLA-4 Through DOX-Mediated Upregulation of FcγRIIIA

Given the enhancement of FcγRIIIA expression after DOX/aPD-1 treatment in GBM patients, we investigated whether this upregulation was due to DOX exposure. Initially, we measured DOX concentration in tumors 2 days postadministration with LIPU/MB, finding levels between 0.25 and 3.95 μmol/kg (Figure 6A). We subsequently assessed the impact of DOX on FCGR3A expression in the HMC3 human microglial cell line, which also expresses the CD163 myeloid marker (Supplementary Figure 13). Exposing HMC3 cells to DOX for 5 hours followed by a 72-hour fresh media incubation resulted in a concentration-dependent increase in FCGR3A transcript levels, up to nearly 16 times that of untreated cells, which surpasses effects seen with IFN-γ exposure (Figure 6C). This increase was confirmed at the protein level by flow cytometry (Figures 6D and E). The presence of DOX presence, indicated by its autofluorescence, persisted in cells for at least 72 hours post-exposure (Supplementary Figure 14). FcγRIIIA upregulation occurred even at low DOX concentrations (0.1 μM), within the range detected in patient tumors (Figure 6A and D). Further testing in primary human myeloid cells derived from PBMC confirmed DOX-mediated concentration-dependent upregulation of FcγRIIIA (Supplementary Figure 15). Moreover, we tested balstilimab (BAL, aPD-1) whether induce FcγRIIIA expression in HMC cells; however, aPD-1 alone did not induce FcγRIIIA expression in HMC3 cells in vitro (Supplementary Figure 16).

Figure 6.

Enhanced efficacy of Fc-enhanced anti-CTLA-4 through upregulation of FcγRIIIA. (A) The DOX concentration in glioblastoma from a patient after 2 days of DOX infusion measured by mass spectrometry. (B) Schematic illustration of in vitro DOX effect assay on human microglial cell line (HMC3). (C) Quantitative RT-PCR analysis of FCGR3A expression upon IFN-γ or DOX treatment. An unpaired t-test was used for statistical analysis. (D and E) Flow cytometry analysis showing FcγRIIIA expression measured as MFI values in HMC3 cells. An unpaired t-test was used for statistical analysis. (F) Kaplan–Meier survival plot of GL261-bearing mice treated with FcE anti-CTLA-4, anti-PD-1/FcE anti-CTLA-4 with and without doxorubicin. (G) Kaplan–Meier curve showing survival of GL261 tumor-bearing mice treated with doxorubicin, anti-PD-1/FcE anti-CTLA-4, and the combination of antibodies and doxorubicin with ultrasound. (H) Scheme of immunophenotyping experimental plan. (I) A pie chart showing proportion of infiltrating immune cells. (J) The CD8-to-CD4 ratio of T cells. (K) Functional state of CD8 T cells analyzed by granzyme B and IFN-γ (left), percentage of granzyme B+ cells in CD8+ T cells (right). (L) The Treg population was assessed by Foxp3+ CD4+ T cells. Percentage of Foxp3+ cells in CD4+ T cells (left) and representative figures are shown. (M) The percentage of IFN-γ+ cells in CD4+ T cells. An ordinary 1-way ANOVA was used for statistical analysis. Data indicate mean ± SD and the P value is depicted.

Next, we investigated the efficacy of the combination strategy in GL261-bearing mice, a model less responsive to FcE-aCTLA-4 compared to QPP4 and CT-2A models (Figure 4B–D). FcE-aCTLA-4 combined with aPD-1 improved survival over monotherapy of FcE-aCTLA-4, achieving 14.3% long-term survival (120 days postimplantation, P = .0021). Survival rates further increased to 85.7% when DOX was added to the regimen (P = .0002; Figure 6F). Notably, integrating LIPU/MB into the treatment doubled long-term survival rates to 33.3% compared to without LIPU/MB (Figure 6F and G). Similarly, treatment with LIPU/MB + IC + DOX resulted in 16.7% long-term survivors (Figure 6G). The most significant survival benefit was observed following LIPU/MB-enhanced delivery of FcE-aCTLA-4, aPD-1, and DOX, leading to 90.9% of long-term survival (Figure 6G). The QPP4 model showed similar results (Supplementary Figure 17).

We characterized the phenotype of immune cells in GL261 tumors treated with this combinatorial immunotherapy (Figure 6H). Immune cells were categorized into lymphocyte (CD45^high CD11b^low), myeloid cell (CD45^high CD11b^high), and microglia (CD45^mid CD11b^high) (Supplementary Figure 5 and Figure 6I). Compared to the IC-treated group, mice receiving FcE-aCTLA-4/aPD-1 showed increased infiltrating lymphocytes, whereas the DOX-only-treated group showed fewer infiltrating immune cells (Figure 6I). To explore the T cell compartment, we assessed the CD8 to CD4 ratio, observing a substantial increase in CD8⁺ T cells for the combination therapy group (Figure 6J). Regarding killer lymphocyte functionality, CD8⁺ T cells showed significant activation, as indicated by the accumulation of granzyme B, although IFN-γ levels remained unchanged (Supplementary Figure 18A and Figures 6K). Alongside this T cell activation, APCs became more favorable to T cell engagement, displaying an increased proportion and enhanced expression of CD86 costimulatory molecule, though CD80 expression did not change significantly (Supplementary Figures 18B–D). NK cell functionality also increased in the treatment groups, marked by elevated IFN-γ and granzyme B (Supplementary Figures 18E–H). FcE-aCTLA-4/aPD-1 therapy, with or without DOX, reduced Treg numbers (Figure 6L) and enhanced IFN-γ expression in CD4⁺ T cells (Figure 6M), confirming the effectiveness of therapy in modifying the tumor microenvironment towards an anti-tumor state.

Discussion

Botensilimab, which is currently under clinical investigation in advanced solid tumors (NCT05529316, NCT04121676, NCT05630183, NCT05608044, NCT05377528, NCT03860272, NCT04028063, NCT05672316, and NCT05627635), is designed to enhance binding to human FcγRs, particularly FcγRIIIA, which has been shown to be critical for the activity of aCTLA-4 antibodies.^21,51 Here, we used a mouse analog of botensilimab, FcE-aCTLA-4, engineered with the same DLE mutations in the Fc region to enhance binding to FcγRIV, the mouse ortholog of human FcγRIIIA. Antibody co-engagement of FcγRs can either activate or inhibit immune responses and compared to its unmodified mouse IgG2b aCTLA-4 variant, the FcE-aCTLA-4 had a significantly higher A/I ratio, a measure previously shown to be predictive of cytotoxicity in vivo.⁴³ We demonstrate that the FcE-aCTLA-4 promoted superior anti-tumor activity than a conventional aCTLA-4 antibody in 3 distinct immunotherapy-resistant murine orthotopic glioma models. The enhanced anti-tumor activity that we observed by FcE-aCTLA-4 antibody was accompanied by a reduction of intratumoral Tregs, decreased myeloid-to-lymphocyte ratio, and increased CD8-to-CD4 ratio. Consistent with the ability of ICB to reinvigorate dysfunctional TILs and augment their antitumor effects,^52,53 our data also show that FcE-aCTLA-4 therapy enhanced the activation and cytotoxic capacity of the tumor-infiltrating CD8⁺ T cells, which were characterized by elevated IFN-γ expression and reduced PD-1 expression.

Given that the FcγRIIIA V158F single-nucleotide polymorphism has been linked to better clinical outcomes with ipilimumab in inflamed tumors,⁵¹ our data suggest that an FcE-aCTLA-4 antibody may offer a wider therapeutic range and effectiveness regardless of FcγRIIIA allele status. This is due to the enhanced binding to both variants, as shown by phase 1 clinical studies for botensilimab.^27,54 Furthermore, our study demonstrates that TAMs in both human and mouse gliomas express FcγRIIIA and FcγRIV, respectively, and CTLA-4 expression is notably higher in intratumoral Tregs. Interestingly, the suppressive effects of Tregs and TAMs in glioma may not be mutually exclusive, as studies in GBM suggest that CTLA-4-mediated immune suppression could be related to the infiltration of macrophages within the tumor microenvironment.⁵⁵ In addition to decreasing the myeloid-to-lymphocyte ratio in mouse models of glioma, our data suggest that the FcE-aCTLA-4, via its increased binding to mouse FcγRIV, is optimized to mediate potent cross-linking of CTLA-4-expressing intratumoral Tregs and FcγRIV-expressing phagocytic TAMs to deplete intratumoral Tregs. This activity was not seen with the unmodified aCTLA-4 antibody and may explain why conventional aCTLA-4 antibodies, ipilimumab, or tremelimumab, have shown limited clinical activity in patients with GBM.^55,56

However, monotherapy responses and long-term survival following FcE-aCTLA-4 treatment differed among glioma models. Notably, GL26 tumor-bearing mice treated with FcE-aCTLA-4 monotherapy succumbed to their disease, suggesting that combination therapy may be needed to overcome super immune-resistant microenvironment of GBM. Many studies have been showing the enhanced delivery effect of LIPU/MB, a procedure that temporarily opens the BBB for impermeable anticancer drug such as immune checkpoint blockade and chemotherapeutics including DOX.^{30–32,34,36} DOX at low doses has been shown to potentiate the response to aPD-1 in breast cancer,³⁸ and with LIPU/MB-mediated delivery, it accumulated more in the human brain and potentiated immune response when combined with aPD-1.³⁶ The combination effect of DOX and aPD-1 was not only on T cells but also on microglia so that those microglia produced more IFN-γ with potentially upregulated antigen-presenting machinery.³⁶ We adapt this combinatorial strategy to address both suppressive immune microenvironment and limited delivery of anticancer drugs. While the combination of FcE-aCTLA-4 and aPD-1 modestly improved long-term survival in GL261 model, combining FcE-aCTLA-4 with aPD-1, DOX, and LIPU/MB resulted in durable anti-tumor response in approximately 90% of treated mice. We hypothesize that this significantly improved response could be attributed not only to increase drug concentration mediated by LIPU/MB but also to the potential that DOX upregulates the expression of activating FcγRs on TAMs and promotes immunogenic cell death, stimulating an immune response that synergizes with ICB therapy.⁵⁷ To support this, we present evidence that in patients with recurrent GBM, the systemic administration of 30 mg of DOX with LIPU/MB leads to an upregulation of FcγRIIIA expression in TAMs. Additionally, exposure to DOX at concentrations observed within human tumors leads to increased FcγRIIIA expression in a human microglia cell line. Notably, our data suggest that DOX could be leveraged to enhance the response to aCTLA-4 by upregulating FcγRs, which are critical for its activity.^21,51,58 Our results suggest that administration of DOX might lead to improvement in the efficacy of other antibody-related forms of immunotherapy where the Fc region binding FcγR in immune cells is important for their mechanism of action.

In conclusion, our study highlights a novel therapeutic approach for the treatment of glioma using an FcE-aCTLA-4 that promotes superior immune activation and durable anti-tumor responses than conventional ICB. This effect was associated with the depletion of intratumoral Tregs by TAMs and enhanced CD8⁺ T cell anti-tumor responses. Moreover, we demonstrate that the expression of FcγRIIIA in TAMs, a receptor that was previously established to be critical for the activity of aCTLA-4 therapy,²¹ can be increased by DOX. Notably, the therapeutic efficacy of FcE-aCTLA-4 is significantly enhanced when combined with DOX, aPD-1, and LIPU/MB while maintaining tolerability. Our results indicate that FcγRIIIA expression is highest in newly diagnosed GBM, suggesting this is the ideal disease state to investigate the efficacy of this immunotherapy approach for these tumors. Collectively, these results and the promising efficacy of botensilimab (FcE-aCTLA-4) observed in ongoing clinical trials in other solid tumors provide the rationale for investigating this novel immunotherapy strategy with LIPU/MB in a clinical trial where this combinatorial strategy is being tested in the adjuvant setting in GBM patients (clinicaltrials.gov NCT05864534).

Supplementary material

Supplementary material is available online at Neuro-Oncology (https://academic.oup.com/neuro-oncology).

Funding

This work was supported by the NIH grants 1R01NS110703-01A1 (A.M.S.), NIH 1R01CA245969-01A1 (A.M.S. and R.St.), 1U19CA264338-01 (A.M.S., R.St., B.Z., C.L.C., J.P., N.B.), P50CA221747 SPORE for Translational Approaches to Brain Cancer (A.M.S., R.St., A.B.H., C.L.C., J.M.), SPORE P50CA097257 (J.P., N.B.), R01 CA120813, RO1 NS120547 (A.B.H.) as well as generous philanthropic support from the Moceri Family Foundation and the Panattoni family. R35CA253126 and Vagelos Precision Medicine award (R.R. and J.Z.). Lunaphore COMET multiplex immunohistochemistry was enabled by a gracious gift from the Stephen M. Coffman trust to the Northwestern Medicine Malnati Brain Tumor Institute of the Lurie Cancer Center. Schematic illustrations are created with BioRender.com.

Conflict of interest statement

A.M.S. and R.St. have received in-kind and/or funding support for research from Agenus, BMS, and CarThera. A.M.S., V.A.A., K.S.K., C.A., and R.St. are co-authors of IP filed by Northwestern University related to the content of this manuscript. A.M.S. is a paid consultant for Carthera and Enclear Therapies. R.St. has acted or is acting as a scientific advisor or has served on advisory boards for the following companies: Alpheus Medical, AstraZeneca, Boston Scientific, CarThera, Celularity, GT Medical, Insightec, Lockwood (BlackDiamond), Northwest Biotherapeutics, Novocure, Inc., Syneos Health (Boston Biomedical), TriAct Therapeutics, Varian Medical Systems. M.C. is an employee and holds ownership interest in Carthera, as well as patents related to the ultrasound technology described herein. R.R. is a founder and a member of the Avisory Board of Genotwin, Diatech Pharmacogenetics, and a consultant for Arquimea Research. None of these activities are related to the work described in this manuscript. D.C., B.M.M., and D.L. are employees of Agenus Bio. All other authors declare that they have no competing interests.

Authorship statement

K.S.K., R.St., and A.M.S. conceptualized the project. K.S.K., K.H., and A.M.S. drafted the manuscript. K.S.K., C.L.-C., D.C., R.St., and M.C. performed the majority of the editing of the manuscript. The experiments presented were conducted and analyzed by K.S.K., K.H., A.G., V.A.A., and C.D. L.C., D.Z., B.C., L.B., D.L., B.M.M., P.Z., J.M., J.Z., R.S., and R.R. performed the bioinformatic analysis; K.S.K., V.A.A., A.M.S., H.N., K.H., K.M., C.H., A.B.H., J.J.P., S.P., and B.Z. performed the histological analysis of the human tumors. N.B. provided samples for analysis and administrative/leadership support. Part of the U19 GTN program for this project. A.M.S. performed the surgeries and implant of SC9 for all patients. M.Mu. performed the quantification of doxorubicin in tissue. Single-patient expanded access protocols and related treatments were performed by A.M.S., C.A., R.St., D.C., C.L.-C., and A.M.S. supervised the study.

Data availability

Single-cell RNA-sequencing dataset corresponding to 201 986 single cells from 44 glioma tumor fragments was acquired from the publicly available database hosted on the Broad Institute Single Cell Portal (singlecell.broadinstitute.org) and used for downstream analysis. Mouse single-cell RNA-sequencing data obtained from Miska et al., Science Advances 2021, DOI: 10.1126/sciadv.abc8929. All data from this study are included in this manuscript and Supplementary Materials and are available upon request from the corresponding author.