Quantitative Cell Type Specific Immunopeptidome Analysis of Macrophage and Tumor Co-evolution Reveals Therapeutic MHC-I Peptides in Glioblastoma

Abstract

Immune checkpoint inhibitors (ICI) have shown impressive performance in treating several types of solid tumors. However, they have been ineffective in glioblastoma (GBM), in part due to the immunosuppressive tumor microenvironment (TME) created by GBM-associated macrophages (GAMs). To uncover MHC-I peptide antigens for targeted immunotherapy, we performed cell type specific immunopeptidome analysis on primary macrophages and GBM tumor cells in a co-culture system to profile MHC-I associated antigen presentation at the tumor-macrophage interface. Co-culturing tumor cells and macrophages induced increased presentation of peptides derived from proteins associated with cytokine signaling pathways on macrophages and from proteins associated with the Rho GTPase pathway on GBM tumor cells. In vivo expression was validated for a cohort of co-culture-induced GAM or GBM associated peptides selected as potential immunotherapy targets, and an mRNA vaccine was developed encoding six peptides from GAMs and GBM tumor cells. Two doses of vaccination generated an antigen specific immune response, significantly delayed GBM tumor growth, and in some cases eradicated tumors. These results demonstrate the translational potential of co-culture induced MHC peptide antigens as therapeutic targets for GBM/GAM targeting vaccines.

Introduction

Glioblastoma is an aggressive form of adult brain cancer with poor overall survival (1). Despite success of immune checkpoint inhibitors and other immunotherapies in treating multiple other solid tumors, they have had limited efficacy in GBM (2–4). This lack of treatment efficacy is associated with an immunosuppressive tumor microenvironment (TME), where complex interactions between tumor cells and the TME collaboratively suppress therapeutic response. Macrophages are the most abundant infiltrated immune cell in GBM, consisting up to 30% of tumor mass(5,6). Recruited and modulated by glioma cells, glioma associated macrophages (GAMs) exert immune suppressive functions in the TME by secreting cytokines to support tumor aggression and inhibit T cell function, upregulating direct receptor-ligand interactions that suppress cytotoxic T cells, and inducing proliferation of CD4 regulatory T cells(7–9). Previous attempts at inhibiting GAMs by targeting cellular receptors including CSF1R, CD73, and CCR2 have shown reduced tumor growth in preclinical models(10–12). Although yet to show clinical efficacy, targeting GAMs to deplete or re-educate the macrophages remains an attractive therapeutic strategy and holds promise to revert the immunosuppressive TME and reinvigorate immunotherapies.

Recognition of tumor associated antigens through TCR-peptide MHC (pMHC) interactions is central to the adaptive immune response against cancer. Immunotherapies that redirect T cells to target tumor associated or tumor specific MHC-I peptides have shown efficacy in melanoma, lung, and breast cancer(13–16). Immunopeptidomic profiling by liquid chromatography tandem mass spectrometry (LC-MS/MS) is an unbiased method to profile MHC-I or MHC-II associated antigens presented on the cell surface that can inform putative therapeutic targets(17,18). Cancer vaccines targeting glioma associated antigens identified by immunopeptidome profiling have been developed and tested in clinical trials(19). For instance, in the Glioma Actively Personalized Vaccine Consortium (GAPVAC), patients treated with a premanufactured library of unmutated antigens elicited a sustained response of central memory CD8+ T cells(20). Another study targeting GBM neoantigens showed that patients who received neoantigen targeting vaccines generated polyfunctional neoantigen specific CD4+ and CD8+ cells(21). To date, however, the MHC-I antigens presented by GAMs that make up a substantial portion of GBM tumor have not been characterized. We reasoned that profiling the antigen presentation landscape at the GBM:macrophage interface could identify targets for GAMs and GBM tumor cells to inform targeted therapies that boost anti-tumor response. Conventional immunopeptidomics studies for antigen discovery profile tumor specimens in bulk, where different cell types may have distinctly different antigen presentation landscapes(22), rendering it challenging to identify cell-of-origin for identified antigens. Resolving cell type specific antigen presentation would assist in accurate target selection to direct T cell response to intended cell populations. To minimize off-target effects on healthy cells, the tumor associated antigen also needs to be expressed at elevated levels in TME, requiring quantitation of antigen presentation on GAM or GBM tumor cells during naïve to adapted state transitions in the TME. Towards this goal, we defined the alterations of GAM and GBM tumor cell immunopeptidomes as they co-evolved in a co-culture model mimicking tumor:immune interactions in the GBM TME to identify targetable GAM or GBM associated MHC-I peptides with therapeutic potential.

Materials and methods

Animals and human subject

C57BL/6 female mice were purchased from Jackson Laboratory (RRID:IMSR_JAX:000664). Mice 6–12 weeks old were acclimated for 3 days before any experiments. All animal procedures were approved by the Committee on Animal Care at MIT. Studies involving human subjects were approved under MSK IRB #09-156. Written informed consent was obtained from patients. Studies were conducted in accordance with Belmont Report.

Cell culture

GL261 cells (RRID:CVCL_Y003) and CT2A cells (RRID:CVCL_ZJ44) were purchased from ATCC in 2020, authenticated using Short Tandem Repeat (STR) profiling. No re-authentication has been performed. Cells were grown in DMEM/F12 medium with 10% FBS at 37C, 5% CO2. Passage number was kept under 15 for all cells used in experiments. Cells were checked every 6 months for mycoplasma using MycoAlert^® Mycoplasma Detection Kit (Lonza). Bone marrow cells were harvested from femurs of C57BL/6 mouse. Mice were euthanized using isoflurane and secondary cervical dislocation. Femurs from the hind limb were dissected and transferred to PBS on ice. The femurs were disinfected with 70% ethanol and both tips were excised. Bone marrow cells were flushed into a 50ml conical tube with ice cold PBS using a 25-gauge needle attached to a syringe. After harvest, bone marrow cells were centrifuged at 300xg for 10 mins and red blood cells were lysed using RBC lysis buffer (Thermo Fisher) for 5 minutes at room temperature. Resulting bone marrow cells were frozen in 90% FBS with 10% DMSO (Sigma) until required for experiments. Bone marrow derived macrophages (BMDMs) were differentiated in complete DMEM media with 50ng/ml mouse M-CSF (PeproTech) for 5 days before co-culture experiments. M-CSF was replenished at Day 3.

Polarization to canonical macrophages

Differentiated BMDMs were polarized to canonical pro-inflammatory macrophages using 100ng/ml IFNγ (Peprotech) and 10ng/ml LPS (Sigma) and to canonical anti-inflammatory macrophages using 40ng/ml IL-4 (Sigma) for 48hrs. Polarized macrophages were lifted with 10mM EDTA (Sigma) in PBS for 10–15 mins. For flow cytometry analysis, cells were washed with FACs buffer for staining. For immunopeptidomics analysis, cells were washed once in PBS before being frozen down in pellet at −80C until peptide MHC isolation.

Co-culture with tumor cells and processing for flow cytometry and immunopeptidomics analysis

For flow cytometry analysis, BMDMs were differentiated in 6 well plate at 3e5 per well. Post differentiation, 1e5 GL261 or CT2A tumor cells were added to differentiated BMDMs. For immunopeptidomics analysis, 10e6 BMDM cells were differentiated in 15cm plates. Post BMDM differentiation on Day 6, 2e6 CT2A or GL261 tumor cells (Wt or H2 null) were added to differentiated macrophages. Plate was swirled to ensure complete cell mixing. Co-culture continued for 72hrs before lifting the tumor cells with Trypsin at 37°C and then separately with 10mM EDTA since 10mM EDTA is toxic to CT2A cells. Cell mixtures for each biological replicate were processed separately. Starting ratios of 1:3 and 1:5 tumor cells to macrophage ratio for flow cytometry analysis and immunopeptidomics analysis were chosen so that the end ratio of tumor: macrophage was consistent between the two experiments. Cell seeding density was chosen so that at the end of co-culture, tumor cells and macrophages were not overcrowded in the culture dish.

Generation of H2-Kb, Db knockout cell lines

H2Kb and H2Db were sequentially knocked out using Crispr-Cas9. Guide RNA was optimized to maximize knockout efficiency. Guide RNA targeting the signal sequence or exon 1 region of H2-Kb or H2-Db was cloned into a DNA Cas9 scaffold acceptor plasmid driven by U6 promoter using restriction site digestion and ligation. Cells were co-transfected with pSpCas9(BB)-2A-GFP (PX458) (Addgene #48138) and guide RNA plasmid using Lipofectamine^™ 3000 Transfection Reagent (Invitrogen #L3000008). Transfected cells were grown for 72hrs and stained with PE anti-mouse H2Db antibody (Biolegend #111507, RRID:AB_313512) and Pacific Blue^™ anti-mouse H-2Kb Antibody (Biolegend #116514, RRID:AB_1967129). Single cell clones with double negative signal were sorted into 96 well plate and grown to form H2 KO CT2A and GL261 tumor cell lines. In case residual expression remained, transfection and sorting were repeated on selected clones.

Flow cytometry

For flow cytometry staining, lifted cells from culture plates were first resuspended in FACS buffer (PBS (Corning) with 5% FBS (Thermo Scientific)). Cells were blocked with TruStain FcX^™ PLUS (anti-mouse CD16/32) Antibody (Biolgend #156603, RRID:AB_2783137) diluted in FACS buffer for 10min on ice to prevent non-specific binding. Cells were then washed with FACS buffer and stained for surface proteins using a cocktail of fluorophore-conjugated antibodies (Supplementary Table 1) resuspended in FACS buffer. When macrophages were present, True-Stain Monocyte Blocker^™ (Biolegend #426103) was added to block Fc binding. Following surface staining, cells were washed twice with FACS buffer and stained with LIVE/DEAD^™ Fixable Violet Dead Cell Stain (Thermo #L34963) diluted in PBS for 15min. Finally, the cells were washed with FACS buffer once before acquisition on flow cytometer. Sample acquisition was performed on BD flow cytometers (FACS Celesta, Symphony A3) and analyzed using FlowJo v10 (TreeStar). For co-culture analysis, macrophages were pre-gated on singlets, live, and CD45+ while tumor cells were pre-gated on singlets, live, and CD45−.

Macrophage and GBM cell separation from co-culture

Harvested BMDM:GBM cell mixture was kept on ice and immediately proceeded to magnetic bead enrichment. We used EasySep^™ Mouse CD45 Positive Selection Kit (STEMCELL # 18945) to perform enrichment according to the manufacturer’s protocol. Briefly, cell mixture was rinsed once with wash buffer (2%FBS/PBS with 1mM EDTA). Cells were counted and resuspended at 1e8/ml in wash buffer and labeled with anti-CD45 antibody for 15 minutes. Magnetic beads were added to labeled cells and incubated for 5min. CD45+ cells were separated on an EasyEights^™ EasySep^™ Magnet (STEMCELL #18103) and unbound cells (CD45−) were washed twice and collected in a 15ml conical tube. The CD45+ cells were washed for a third time (discard wash) before taking off the magnetic beads and collected into a 15ml conical tube. An aliquot of CD45+ and CD45− cells from each co-culture experiment was reserved for flow cytometry quality control analysis while the rest was frozen in pellet at −80°C for immunopeptidomic analysis.

Peptide synthesis

Heavy amino acid labeled peptides were generated at the Biopolymers & Proteomics core facility at MIT. Synthesized peptides were cleaved using standard cleavage cocktail and purified to >95% using HPLC. Molecular weight was confirmed using MALDI mass spectrometer (Bruker microflex). Heavy isotope labeled amino acids used for synthesis were purchased from Cambridge Isotope Laboratories, Inc.

Generation of recombinant heavy isotope-labeled peptide MHCs (hipMHCs)

Heavy amino acid labeled peptides and positive control peptides (provided by manufacturer) were loaded on recombinant empty mouse H2-Kb or H2-Db monomers, Kb/Db easYmers^®, from Immunaware according to manufacturer’s protocol. The concentration of stable complexes post loading was quantified using adapted protocol from Flex-T HLA class I ELISA assay (Biolegend). Briefly, positive control peptide and heavy peptide loaded complexes in serial dilution were incubated in streptavidin coated plates. Stable complex was detected using HRP conjugated antibody against β2m protein (Biolegend #280303) and visualized by adding substrate solution. Results were acquired using a Tecan plate reader Infinite 200 with Tecan icontrol version 2.0.0.0. Generated hipMHCs were frozen in aliquots at −20°C until use.

Peptide MHC (pMHC) isolation and multiplexing

Cell pellets of separated BMDMs and GBM tumor cells were resuspended in 1 mL MHC lysis buffer [20 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.2 mM PMSF (Sigma), 1% CHAPS (Sigma), and 1x HALT Protease/Phosphatase Inhibitor Cocktail (Thermo Scientific)], followed by brief sonication at 4 °C (3 × 10 second microtip sonicator pulses at 30% amplitude) to disrupt cell membranes. Lysate was cleared by centrifugation at 16,000 g for 15 minutes at 4°C. Peptide MHCs were isolated from the lysates by immunoprecipitation (IP) and size exclusion filtration, as previously described(23). Briefly, for each sample 0.1 mg of anti-mouse H2-Kb antibody (Y-3 clone, InVivoMAb #BE0172, RRID:AB_10949300) and 0.05mg of anti-mouse H2-Db antibody (produced in house), or 0.5mg anti-human MHC Class I antibody (W6/32 clone, InVivoMab #BE0079, RRID:AB_1107730) were conjugated to 20 μL FastFlow Protein A Sepharose bead slurry (Cytiva) for 4 hours rotating at 4 °C. Beads were washed 1x with IP wash buffer (20 mM Tris-HCl pH 8.0, 150 mM NaCl), followed by addition of lysate and if applicable, 100fmol of each hipMHC, and incubated rotating overnight at 4 °C to immobilize pMHCs to beads. Beads were washed with 1x TBS and 2x HPLC grade water, and pMHCs were eluted using 10% acetic acid for 20 minutes at room temperature (RT). Peptides were isolated from antibody and MHC molecules using a 10K molecule weight cutoff filter (PALL Life Science), lyophilized, and stored at −80 °C until analysis.

For immunopeptidome characterization of GAM and adapted GBM tumor cell comparison to the naïve state, isolated pMHCs from macrophage and tumor cells in the co-cocultured state and naïve state were labeled with TMTpro^™ 18-plex Label Reagent (ThermoFisher #A52045) according to manufacturer’s protocol. Briefly, lyophilized pMHCs were resuspended in labeling buffer (50% Ethanol, 35% HPLC water, 15% TEAB) and labeled with TMTpro reagent resuspended in anhydrous acetonitrile. After 3.5hr labeling at room temperature, the reaction was quenched by 5% hydroxylamine (ThermoFisher #90115). Labeled samples were pooled, dried using speedvac, and stored at −80 °C until analysis.

Sample processing for analysis of macrophage and GBM proteome

Lysate was collected post-IP of MHC-I complex from GAMs, adapted GBM cells, and their naïve state. Bicinchoninic Acid Assay (Pierce^™ # 23225) was used to determine the concentration of each sample. 100μg protein from each sample was processed through S-Trap^™ sample processing technology according to manufacturer’s protocol (Protifi). Briefly, lysate was diluted in SDS, reduced using DTT (Sigma # D9779-25g) and alkylated using IAA (Sigma # I1149-5G). Protein was trapped in mini or micro S-Trap column and cleaned using wash buffer (100mM TEAB in methanol). Digestion was performed with trypsin at 47°C for 1hr. Peptides were collected from the column and lyophilized for analysis.

To compare the proteome of macrophages or tumors cells in the co-cultured state to naïve state, TMTpro reagent was used to multiplex the samples. Samples were resuspended in 50mM HEPEs (pH 8) and labeled according to manufacturer’s protocol (described in the previous section).

High-pH Reversed-Phase Liquid Chromatography (RPLC) fractionation

Fractionation was performed on an Agilent fractionator (1100 series) with Buffer A: 10 mM TEAB in water (dilute from the 1 M TEAB stock solution) and Buffer B: 10 mM TEAB in 90% Acetonitrile.

Fractionation of pMHC samples: TMT labeled pMHC samples or unlabeled pMHC samples were resuspended in 25μl 5% Buffer B in Buffer A and loaded onto a 20 cm fractionation capillary chromatography column prepared and packed in house (200 μm ID & 10 μM C18 beads, ReproSil-Pur) at 50–100 psi pressure. For TMT labeled samples the fractionation column was washed with 5% Buffer B for 30min before fractionation. Samples were fractionated at flow rate of 0.3ml/min using a gradient with increasing % of Buffer B: 0–3% for 0.5min; 3–40% for 54.5min; 40–100% for 3min; 100–0% for 2min. Fractionation column was manually moved among 6 screw cap tubes in TMT labeled co-culture pMHC analysis and 10 tubes in unlabeled human macrophage:GBM pMHC analysis (2min each) to result in 6 total fractions. The elution was lyophilized and subjected to MS acquisition.

Fractionation of global proteome samples: TMT labeled proteome samples were resuspended in 200μl 5% Buffer B and loaded onto a Kromasil^® C18 HPLC Column (Sigma, 5 μm particle size, pore size 100 Å, L × I.D. 250 mm × 4.6 mm). Samples fractionated at flow rate of 1ml/min using a gradient of increasing % of Buffer B: 0–1% for 1min; 1–5% for 4min; 5–40% for 60min; 40–70% for 10min; 70% for 9min; 70–1% for 1min on an automated fractionator into 10 fractions. Ten percent of the fraction was aliquoted and lyophilized to perform MS acquisition.

Mass spectrometry data acquisition

Samples were analyzed using an Orbitrap Exploris 480 mass spectrometer (Thermo Scientific) coupled to an UltiMate 3000 RSLC Nano LC system (Dionex), Nanospray Flex ion source (Thermo Scientific), and column oven heater (Sonation). The peptide MHC sample was resuspended in 5μl of 3% acetonitrile, 0.1% Formic Acid, and loaded onto a 10–12 cm analytical capillary chromatography column with an integrated electrospray tip (~1 μm orifice), prepared and packed in house (50 μm ID & 1.9 μM C18 beads, ReproSil-Pur) through WPS-3000 autosampler (Dionex).

DDA analyses of TMT labeled pMHC samples: Peptides were eluted using a gradient with 3–10% buffer B (70% Acetonitrile, 0.1% formic acid) for 5 minutes, 10–30% for 65 minutes, 30–35% for 7 minutes, 35–55% for 5 minutes, 55–97% for 10 minutes, and 97% to 3% for 2 minutes. Standard mass spectrometry parameters were as follows: spray voltage, 2.5 kV; no sheath or auxiliary gas flow; heated capillary temperature, 275 °C. Full scan mass spectra (300–1200 m/z, 120,000 resolution) were detected in the orbitrap analyzer after accumulation of 3e6 ions (normalized AGC target of 300%). For every full scan, up to 3sec cycle time of ions were subsequently isolated if the ions reached a minimum intensity threshold of 5e3 and has charge state of 2–4. For MS2 acquisition, ions were collected with an isolation window of 0.4 m/z, 120,000 resolution, maximum injection time of 250ms, normalized AGC target = 110%, and fragmented by higher energy collisional dissociation (HCD) with a collision energy (CE): 33%. Ions were excluded for 30s after being acquired 2 times within 20s.

DDA analyses of unlabeled pMHC samples: Peptides were eluted using a gradient with 6–15% buffer for 13 minutes, 15–30% for 35 minutes, 30–45% for 20 minutes, 45–55% for 8 minutes, 55–97% for 2 minutes, and 97% to 3% for 1 minute. Standard mass spectrometry parameters were as follows: spray voltage, 2.5 kV; no sheath or auxiliary gas flow; heated capillary temperature, 280 °C. Full scan mass spectra (300–1200 m/z, 60,000 resolution) were detected in the orbitrap analyzer after accumulation of 3e6 ions (normalized AGC target of 300%). For every full scan, up to 3sec cycle time of ions were subsequently isolated if the ions reached a minimum intensity threshold of 5e3 and has charge state of 2–4. For MS2 acquisition, ions were collected with an isolation window of 0.4 m/z, 120,000 resolution, maximum injection time of 250ms, normalized AGC target = 1000%, and fragmented by higher energy collisional dissociation (HCD) with a collision energy (CE): 30%. Ions were excluded for 30s after being acquired 3 times within 20s.

DDA analyses of TMT labeled proteome samples: Peptides were eluted using a gradient with 6–19% buffer B for 38 minutes, 19–29% for 17 minutes, 29–41% for 9 minutes, 41–97% for 3 minutes, and 97% to 3% for 1 minute. Standard mass spectrometry parameters were as follows: spray voltage, 2.5 kV; no sheath or auxiliary gas flow; heated capillary temperature, 275 °C. Full scan mass spectra (380–2000 m/z, 60,000 resolution) were detected in the orbitrap analyzer after accumulation of 3e6 ions (normalized AGC target of 300%) or 25 ms. For every full scan, up to 3sec cycle time of ions were subsequently isolated if the ions reached a minimum intensity threshold of 1e4 and has charge state of 2–6. For MS2 acquisition, ions were collected with an isolation window of 0.4 m/z, 120,000 resolution, maximum injection time of 250ms, normalized AGC target = 100%, and fragmented by higher energy collisional dissociation (HCD) with a collision energy (CE): 33%. Ions were excluded for 30s after being acquired 2 times within 25s.

SureQuant Survey analyses: Peptides were eluted using a gradient with 6–25% buffer B (70% Acetonitrile, 0.1% formic acid) for 53 minutes, 25–45% for 12 minutes, 45–97% for 3 minutes, and 97% to 3% for 1 minute. Standard mass spectrometry parameters were as follows: spray voltage, 2.5 kV; no sheath or auxiliary gas flow; heated capillary temperature, 280 °C. The Exploris was operated in data dependent acquisition (DDA) mode with an inclusion list of the heavy trigger peptide. Full scan mass spectra (300–1500 m/z, 60,000 resolution) were detected in the orbitrap analyzer after accumulation of 3e6 ions (normalized AGC target of 300%) or 50 ms. For every full scan, up to 20 ions were subsequently isolated if the m/z was within +/− 5 ppm of targeted trigger peptide and reached a minimum intensity threshold of 1e5. Ions were collected with a maximum injection time of 250ms, normalized AGC target = 1000%, and fragmented by higher energy collisional dissociation (HCD) with a collision energy (CE): 30%. Library of acquired spectra was generated using Skyline software for SureQuant targeted analyses.

SureQuant targeted analyses: The custom SureQuant acquisition template available in Thermo Orbitrap Exploris Series 2.0 was utilized for this method build. All the acquisition parameters for heavy labeled peptides are located within a distinct 4-node branch stemming from a full scan node. Each branch houses information of peptides with the same heavy labeled amino acid. In the full scan, the trigger peptide m/z and intensity thresholds are defined in the “Targeted Mass” filter node as 1% of the intensity from DDA survey run. Next, parameters for the low resolution, trigger peptide MS2 scan are defined, followed by the “Targeted Mass Trigger” filter node, which defines the 6 product ions used for pseudo-spectral matching (Supplementary Table 2). To connect each set of product ions within the targeted mass trigger node to a given precursor mass, group ID feature was utilized to define the precursor m/z associated with each group of product ions. Finally, the isolation offset (m/z) corresponding to the MS2 scans of the endogenous peptides was defined in the scan parameters within each node. Standard mass spectrometry parameters for SureQuant acquisition are as follows: spray voltage: 2.5kV, no sheath or auxiliary gas flow, heated capillary temperature: 280°C. Full-scan mass spectra were collected with a scan range: 380–1200 m/z, AGC target value: 300% (3e6), maximum IT: 50 ms, resolution: 120,000. Heavy peptides matching the m/z (within 10 ppm) and exceeding the intensity threshold defined on the inclusion list were isolated [isolation window 1 m/z] and fragmented [nCE: 30%] by HCD with a scan range: 150–1200 m/z, maximum IT: automatically determined from the resolution, AGC target value: 1000% (10e6), resolution: 15,000. A product ion trigger filter next performs pseudo-spectral matching, only triggering an MS2 event of the endogenous target peptide at the defined mass offset if n ≥ 3 product ions are detected from the defined list. If triggered, the subsequent endogenous peptide MS2 scans are initiated at the defined mass offsets. Scan parameters have the same CE, scan range, and AGC target as the heavy trigger peptide, but with a higher maximum injection time and resolution (max IT: 250 ms, resolution, 120,000).

SureQuant data analysis:

Skyline software (RRID:SCR_014080) was used to quantify abundance of the standard and endogenous peptide. For each sample, the abundance of each target peptide (heavy and light) was approximated using an average of the maximum intensity of top 3 product ions across the elution chromatogram. The absolute amount of endogenous light peptide was calculated using the ratio of light:heavy abundance (heavy peptides were spiked in at 100fmol). Copies of endogenous peptide per cell were calculated using the absolute amount and input cell number in the IP.

$$\frac{Copies}{cell}\left(light\right)=\frac{lightabundance}{heavyabundance}\mathrm{*}\frac{6.02e8}{\#cell}$$

Mass spectrometry data search

All mass spectra from DDA MS were analyzed with Proteome Discoverer (PD, version 3.0, RRID:SCR_014477), searched using SEQUEST against the Swiss-Prot mouse database (Identifier: Mus musculus [sp_canonical TaxID=10090_and_subtaxonomies], release 2024_06) or Swiss-Prot human database (Identifier: Homo sapiens [sp_canonical TaxID=9606_and_subtaxonomies], release 2024_06), and rescored using INFERYS. No enzyme was used, variable modifications included oxidized methionine for all analyses, and static modifications included Carbamidomethyl, TMTpro on Lysine and N-terminus. All analyses were filtered with the following criteria: search engine rank = 1, length between 8 and 12 amino acids, q value ≤ 0.05. A separate quality control search using the same parameters was performed using a combined database comprised of the PD Swiss-Prot mouse database and bovine database (Identifier: Mus musculus [sp_canonical TaxID=10090_and_subtaxonomies; Bos taurus (sp_canonical TaxID=9913)]. Any peptide spectrum match (PSM) in the mouse database search that matched to a different bovine protein was discarded.

Mouse CT2A and GL261 intracranial tumor immunopeptidome data was downloaded from ProteomeXchange Consortium via the PRIDE partner repository (PXD036720) and searched using the same criteria described above.

MS data processing

Processed TMT intensities for each experiment are shown in Supplementary Table 7–12.

Immunopeptidome data:

TMT intensities from PSMs matching to the same peptide (including oxidized and non-oxidized form if any) were summed. Peptides with empty values in any channel (sample) were discarded before analysis. Log2FC was calculated by log2 value of co-cultured sample TMT intensity divided by mean intensity of unperturbed samples (n≥3) for each peptide. When analyzing GBM cell immunopeptidome post co-culture, we removed potential macrophage derived peptides from our results using the following filtering steps.

1. Sort peptides by mean TMT intensity of monoculture GBM cells (GL261 or CT2A)

2. Store the bottom 100 peptides as GL261 or CT2A low presentation peptides. These peptides are more prone to be falsely identified as co-culture upregulated peptides if the co-cultured tumor cells have slight macrophage contamination since macrophages generally have higher peptide abundance.

3. If any of the peptides from step 2 was identified as a putative co-culture upregulated peptide, exclude from further analysis.

Global proteome data:

TMT intensities from MS2 scans of each peptide (including oxidized and non-oxidized form if any) were summed. Peptides with empty values in any channel (sample) were discard. The intensities of each sample (channel) were normalized according to the following steps

1. Sum intensities for all peptides in each TMT channel to get TMT channel intensity

2. Calculate the mean of TMT channel intensity and divide each TMT channel intensity by the mean. This is normalization factor for each channel.

3. Divide the individual peptide TMT intensity in each channel by the channel specific normalization factor to get normalized global proteome data.

Log2FC was calculated using normalized TMT intensities.

modRNA production for mRNA vaccine

The mRNA vaccine sequence was designed by linear connection of 6 co-culture upregulated peptide sequences (Supplementary Table 3), where each peptide was separated by AAY linker to ensure proteasomal cleavage (24,25). Template DNA plasmids used in the production of modRNA were created using a commercially available Cloning Kit for mRNA Templates (Takara #6143) according to manufacturer’s instructions. Resultant plasmid DNA was linearized via endonuclease digestion and purified with PureLink PCR Purification columns (ThermoFisher #K310002) following manufacturer’s instructions. To synthesize RNA, 20μL in vitro transcription (IVT) reactions were performed using reagents from the HiScribe T7 High Yield RNA Synthesis Kit (NEB #E2040) and 1–2μg of linear DNA template (scaled as needed). Importantly, modified base N1-methylpseudouridine triphosphate (TriLink #N-1081) was added to the reaction mixture instead of canonical uridine triphosphate, and CleanCap Reagent AG (TriLink #N-7113) was utilized to co-transcriptionally add 5’ Cap-1 structures to synthesized RNA. The IVT product was purified using PureLink RNA Mini columns (ThermoFisher #12183018A) following manufacturer’s instructions. Quality of the resulting modRNA was assessed using UV-Vis spectrophotometry and gel electrophoresis.

Lipid nanoparticle synthesis

9-Heptadecanyl 8-{(2-hydroxyethyl)[6-oxo-6-(undecyloxy)hexyl]amino}octanoate (SM-102) was purchased from BroadPharm (CAT#BP-25499); 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC; CAT#850365), Cholesterol (CAT#700100), 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (DMG-PEG2k; CAT#88015) were purchased from Avanti Polar Lipids. Citrate buffer (pH 3; CAT#J61391-AK) was purchased from Alfa Aesar. For dialysis, 20K MWCO Slide-A-Lyzer^™ MINI Dialysis Device (ThermoFisher Scientific), and RNAse-free PBS (AM9625; ThermoFisher Scientific) were used. Lipid nanoparticles were synthesized using a microfluidic organic-aqueous precipitation method. The organic phase was prepared by mixing the lipids SM-102, DSPC, Cholesterol, and DMG-PEG2k in ethanol at a molar ratio of 50:10:38.5:1.5. The aqueous phase of RNA was prepared by diluting the RNA (stored in RNAse-free water) with 10 mM citrate buffer at pH 3.0. The two phases were prepared at an ethanol:aqueous volume ratio of 1:3, and at an N:P ratio of 5:1. Each phase was loaded into a syringe (BD), and locked onto the NxGen microfluidic cartridge for mixing using a NanoAssemblr Ignite instrument (Precision Nanosystems). The Ignite was set to operate with the following settings: volume ratio- 3:1; flow rate- 12 ml/min; waste volume- 0 mL. The resulting LNPs were dialyzed against 20mM Tris Acetate with 8% sucrose using 10K MWCO Slide-A-Lyzer^™ MINI Dialysis casettes (ThermoFisher Scientific) at 25°C for 4 hours.

Subcutaneous tumor inoculation, vaccination, and monitoring

GL261 tumor cells were cultured for 2 passages before inoculation. Cells were lifted using trypsin and washed with PBS for 2 times. Cells were resuspended to a density of 100e6 cells/ml in PBS and injected on the right flank of C57BL/6 mice at 100μl per mouse (10e6 per mouse). Mice were randomized into different groups post tumor cell injection by another lab member blinded from tumor injection process. In single dose experiment, mRNA vaccine (10μg/dose) was injected intramuscularly on both hindlimbs at Day 8 post tumor inoculation. In double dose experiments, mRNA vaccine (10μg/dose) was injected intramuscularly on both hindlimbs at Day 8 and Day 15 or Day 1 and Day 15 post tumor inoculation. Additionally, mRNA vaccine encoding OT-I SIINFEKL antigen (10μg/dose) was injected at Day 8 and Day 15 as a control for irrelevant antigen control. Tumor size was measured using digital caliper twice per week to monitor tumor growth until the mice reached euthanasia criteria. For ELISPOT measurement, mice were euthanized Day 22 post tumor inoculation.

Tissue processing for ELISPOT, IHC, and Histology

Tumors and spleens were resected and stored in RPMI 1640 media (Gibco) on ice before further processing. Half of spleens were mashed through 70μm filters to obtain a single-cell solution and red blood cells were lysed with RBC lysis buffer (ThermoFisher) for 5min at room temperature. Splenocytes were washed twice with RPMI before further processing. Tumors were flash frozen in liquid nitrogen and stored at −80°C. Portions of tumor and half of spleens were frozen in Tissue-Tek^® O.C.T. Compound (Sakura) on dry ice and sectioned on a cryostat into 5μm sections.

Immunohistochemistry (IHC)

Tumor cryosections were dried at room temperature for 15mins before fixing in 4% paraformaldehyde solution for 15mins. Slides were washed twice with TBST (TBS with 0.05% Tween 20 + 0.025% Triton X-100), each time with 5min incubation. Slides were then incubated with avidin solution for 10mins with biotin solution for 10mins (Biolegend # 927301). 10% BSA was used to block slides for 30min and slides were incubated with primary antibody (CD8 [Invitrogen #MA1–10301] 1:50, RRID:AB_11155388; Granzyme B [CST #44153S], 1:100, RRID:AB_2857976) diluted in 2%BSA/TBS overnight at 4°C. After primary antibody incubation, slides were washed with TBST for 3 times and TBS for 2 times and incubated in 3% H₂O₂ for 15min. Slides were incubated with biotinylated secondary antibody (Jackson ImmunoResearch) for 30min and stained with ABC solution (vectastatin) for 30min. Slides were developed with DAB substrate and counterstained with 20% hematoxylin before mounting for imaging. Quantification of positive signal was performed using Aperio ImageScope (version 12.3.3.5048) positive pixel count algorithm.

Immunofluorescence (IF)

BMDM and CT2A/GL261 co-cultures were prepared in coverslip bottom 6-well plates. Cells were fixed using 4% formaldehyde in PBS for 15mins (Room temperature). Cells were rinsed with PBS and permeabilized with PBS/0.3% Triton^™ X-100 for 10mins. 5% BSA / 0.3% Triton^™ X-100 in PBS was used to block cells for 1hr at room temperature. Cells were stained with primary antibodies (CD45 [Abcam #23910] 1:500, RRID:AB_447758; NF-κB p65[CST #8242S] 1:500, RRID:AB_10859369) diluted in 1% BSA / 0.3% Triton^™ X-100 in PBS overnight. Cells were rinsed and then stained with secondary antibodies diluted in 1% BSA / 0.3% Triton^™ X-100 in PBS for 1.5hrs. Cells were rinsed, and stained with fluorophore conjugated phalloidin ([CST #8878S] 1:100) for 15mins and then 0.5μg/mL DAPI for 5 min. Images were acquired on Leica SP8 Spectral Confocal Microscope at 100x oil objective. Images were analyzed using ImageJ (Version Version 1.54f). Ratio of nuclear vs. cytosol NF-κB intensity was calculated by average intensity of respective segment across a line through the cell. Intensity profile was determined using Analyze→plot profile. Nuclear segment was determined by DAPI stain boundary and cytosol segment was determined by phalloidin stain boundary. Significant difference between untreated BMDM and co-cultured condition was determined by two-tailed unpaired t test (n≥6 cells analyzed).

IFNγ Elispot assay

ELISpot plates (EMD Millipore) were coated overnight at 4°C with anti-IFNγ capture antibody (BD Biosciences, RRID: AB_2868944). Plates were washed and blocked with complete RPMI1640 media for 2h at room temperature. Splenocytes were plated in complete media at 1e6 cells/well with either 10μg/well co-culture upregulated peptide mixture, 10μg/well single co-culture upregulated peptides or OT-I peptide (irrelevant antigen control), negative control (complete media) or positive control (complete media supplemented with 100 ng/mL PMA (Sigma-Aldrich) and 1 mg/mL ionomycin (Sigma-Aldrich)). Plates were incubated overnight (18h) at 37°C and 5% CO2 and developed using a mouse IFNγ ELISpot kit (BD Biosciences # 551083), following manufacturer’s instructions. After drying (overnight at RT), spot counts were determined using an ImmunoSpot (CTL) Elispot reader.

Statistical analysis

Analyses were performed using GraphPad Prism 10 (RRID:SCR_002798), python and R. Comparisons between groups were performed using parametric (student’s t test) or non-parametric tests (Mann-Whitney U test, Kruskal-Wallis test) after F-test. Correction for multiple comparisons was performed where applicable. False discovery rates were calculated using Benjamini-Hochberg method. P values < 0.05 were considered statistically significant (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns = not significant). GibbsCluster 2.0 was used for motif analysis. NetMHCpan 4.1 was used to determine pMHC binding affinity.

Data availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the data set identifier PXD057588. All other raw data generated in this study are available upon request from the corresponding author.

Result

Distinct antigen repertoire presented by naïve, pro-inflammatory and anti-inflammatory macrophages highlights peptides derived from canonical pathways involved in the polarization process

To determine whether macrophage polarization resulted in altered MHC-I presentation, we quantified MHC-I associated peptide presentation of pro-inflammatory (pro) macrophage (IFN-γ+LPS) and anti-inflammatory (anti) macrophages (IL-4) compared to the naïve state using isobaric tandem mass tag (TMT) labeling coupled with mass spectrometry based immunopeptidomic analysis. In line with previous reports(26–28), pro-inflammatory macrophages displayed significantly increased CD80 and MHC-I expression, while anti-inflammatory macrophages upregulated CD206 expression compared to naïve macrophages (Figure 1A). Both polarization processes resulted in alteration of presented peptides, with a larger change in peptide abundance observed in pro compared to anti (Figure 1B).

Figure 1. Immunopeptidomic analysis of pro- and anti-inflammatory macrophages highlights peptides involved in canonical pathways during polarization process.

A). Flow cytometry analysis of pro- and anti-inflammatory markers, and overall MHC-I expression. Data presented as mean ± SD (n=6 per group). Group differences determined using two-tailed student’s t test. B). Log2 Fold Change distribution of MHC-I associated peptide abundance of pro- and anti-inflammatory macrophages compared to naïve macrophages using isobaric tandem mass tag (TMT) labeling. C-D). Heatmap of Log2 FC of profiled MHC-I associated peptides during pro-inflammatory (C) or anti-inflammatory (D) polarization (n=3). Log2FC denotes the Log2 value of pro-/anti-inflammatory macrophage peptide abundance divided by mean abundance of corresponding peptide in naïve BMDM (n=3) E). Overlap of identified peptides and source proteins with increased presentation (p<0.05, Log2FC>0.5) in pro- and anti-inflammatory macrophages. p value calculated using student’s t-test, where the variance difference between two groups is determined by F-test. F). Gene Set Enrichment Analysis of source proteins ranked by log2FC presentation in pro-inflammatory macrophages. Hallmarks gene set from the MSigDB was used. Significant positive or negative enrichment sets with adjusted p<0.05 are shown. A background comprised of all profiled proteins in the immunopeptidome experiment was used for the enrichment analysis. G). Log2FC of selected MHC-I associated peptides from interferon gamma response associated proteins profiled from pro- and anti-inflammatory macrophages. H). Number of peptides identified from interferon gamma pathway associated proteins in pro-inflammatory macrophages. I). Maximum difference in fold change among peptides identified from the same IFN gamma pathway associated source protein in pro-inflammatory macrophages.

Despite overall increased MHC-I expression in proinflammatory macrophages, MHC-I associated peptides show differential changes in abundance, with 716 peptides increased and 221 peptides decreased in presentation by greater than 1.41-fold (0.5 Log2FC) out of 2824 profiled peptides (p<0.05) (Figure 1C, Supplementary Figure 1A, Supplementary Table 6). In contrast, peptides from anti-inflammatory macrophages had limited change in abundance (Figure 1D, Supplementary Figure 1B, Supplementary Table 7), with none of 3553 profiled peptides decreasing in presentation, and only 157 increasing in presentation. Although more than 60% of source proteins (1530) were profiled in both pro- and anti-inflammatory macrophages, less than 10% of the increased source proteins were shared between the two states (Figure 1E, Supplementary Figure 1C). The low fraction of overlapping source proteins demonstrates that pro- and anti-inflammatory macrophage underwent divergent changes in antigen presentation, creating different MHC-I peptide repertoires.

To interpret biological processes that contributed to the presented antigens, we performed gene set enrichment analysis (GSEA) against hallmark gene sets on source proteins ranked by log2FC. Pathways including “interferon gamma response” and “interferon alpha response” showed significant positive enrichment in pro-inflammatory macrophages, reflecting known cellular response to IFNγ and LPS stimulation (Figure 1F). Negative enrichment of G2M checkpoint and E2F targets suggested that known cell cycle regulation during pro-inflammatory polarization also affected antigen presentation(29,30). Other source proteins with more than 1.41 FC (0.5 log2FC) in presentation enriched for pathways related to the innate immune system and cytokine/interleukin signaling, reflecting the regulation by IFNγ (31–33) (Supplementary Figure 1D). More importantly, the profiled immunopeptidome revealed upregulated MHC-I peptides specific to pro-inflammatory macrophages. Peptides of interferon gamma pathway associated proteins such as Cd74, Stat1, Isg15, and Nfkbia increased by up to 4.7 log2FC (26-fold) in pro-inflammatory macrophages whereas in anti-inflammatory macrophages their presentation level remained unchanged (Figure 1G). Although proteins targeted by polarizing cytokines were generally over-represented during polarization, the constituent peptides from these proteins were not equally increased in presentation. For instance, MHC peptides from 22 interferon gamma response pathway associated source proteins increased in presentation in pro-inflammatory macrophages; across these proteins, increased peptide presentation spanned more than 40-fold (Figure 1H–I). As a couple of examples, Parp14, a poly-ADP-ribose polymerase stimulated by IFNγ, had 10 peptides profiled, with a range of 1.23–8 FC during pro-inflammatory polarization (Supplementary Figure 1E). In comparison, the 4 peptides from Ptgs2 spanned from 2.06–48.84 FC. Through this comparison between pro- and anti-inflammatory macrophage, we demonstrate that macrophage antigen repertoire is closely associated with their cellular state, and that direct identification of peptides with upregulated presentation can highlight putative cell surface targets for specific macrophage states.

GBM cell lines polarize BMDM in vitro into a tumor associated state and induce alterations in MHC-I peptide presentation that reflect intracellular response

In the TME, macrophages are recruited and polarized to tumor associated macrophages (TAMs) by tumor cells. These TAMs exert immunosuppressive functions including anti-inflammatory cytokine secretion, promotion of tumor invasiveness, and suppression of effector T cell cytotoxicity(34–36). Targeting TAMs has emerged as a therapeutic effort to revert the immunosuppressive TME and promote an anti-tumor response in multiple cancers, including GBM(11,37). We reasoned that increased presentation of endogenous MHC-I associated antigens in TAMs during interaction with GBM tumor cells could be a source of antigens enabling direct targeting of TAMs in the GBM TME. To model GBM-macrophage interactions with the ultimate goal of identifying TAM associated pMHC antigens or GBM associated pMHC antigens, we used a co-culture system wherein either of two murine GBM cell lines, CT2A or GL261, were co-cultured in physical contact with primary BMDMs. After 72hrs of co-culture, BMDMs displayed a minor increase in phagocytosis activity compared to naïve macrophages, but the overall percentage of phagocytosing cells was less than 8% (Figure 2A). We observed an increase in both commonly used pro- and anti-inflammatory markers CD80 and CD206, validating previous reports that TAMs have a unique phenotype distinct from the canonical pro- or anti-inflammatory axis(38) (Figure 2B–C). Expression of other TAM associated markers CXCR4, PD-1 and CSF-1R (GL261 only) was also significantly increased post co-culture(35,39–43) (Figure 2D–F). Additionally, expression of MHC-I and MHC-II increased post co-culture (Figure 2G–H). These surface marker changes indicate that co-culture with GBM cells resulted in polarization of BMDMs into a GAM-like state distinct from pro- or anti-inflammatory states. Additionally, the two GBM cell lines yielded differential surface marker expression on macrophages, with GL261 co-cultured BMDM showing higher CSF-1R, lower CXCR4 and lower CD206 expression than CT2A co-cultured macrophages. This change in cell surface marker expression could be attributed to different immunogenic properties of the two tumor cell lines, as CT2A cells have been shown to be more immunosuppressive, recruiting more resident macrophages and less T cells in vivo(44).

Figure 2. Macrophages were polarized to tumor-associated state by CT2A/GL261 GBM cells in co-culture.

A). Percentage of tumor cell (CSFE labeled) phagocytosing unperturbed macrophages or macrophages post 72hrs of co-culture with CT2A or GL261 cells. Data presented as mean ± SD (n≥5 per group). Significant difference between unperturbed and co-culture macrophage was determined by two-tailed student’s t-test B-F). Co-cultured macrophage expression of pro-inflammatory associated marker CD80 (B), anti-inflammatory associated marker CD206 (C), TAM associated markers CSF1R (D), PD-1 (E), CXCR4(F), and MHC expression marker MHC-I (G) and MHC-II (H) determined using flow cytometry. Data presented as mean ± SD (n=6 per group). Group differences between co-cultured and unperturbed macrophages determined using two-tailed student’s t test.

We used this established model to quantify alterations in MHC-I associated peptide presentation on GAMs and tumor cells post co-culture, with the goal of identifying peptides with increased presentation that could be used to target the GBM TME. Since tumor cells and macrophages present distinct immunopeptide repertoires, defining the cell-type specific quantitative difference of presented MHC-I associated peptides is important to uncover peptide targets while also providing insight as to the underlying differential biological processes during tumor cell and macrophage co-evolution. To accomplish this goal, differentiated BMDM and GBM cells were analyzed independently or were co-cultured and then separated into macrophage (CD45+) or tumor cells (CD45−) prior to analysis to compare their co-evolved states with their respective naïve states (Figure 3A). We validated that sample processing did not significantly alter the cell immunopeptidome using mock enrichment from single cell-type populations (Supplementary Figure 2A). The selected seeding density of tumor cells and BMDMs resulted in final tumor to macrophage ratio of 7:3 (Supplementary Figure 2B). Purities of macrophage and tumor cell populations after enrichment were both above 90% (Supplementary Figure 2C).

Figure 3. GBM cells alter macrophage MHC-I peptide presentation, inducing putative co-culture upregulated peptides enriching for cytokine signaling pathways.

A). Schematic of cell type specific immunopeptidome analysis in macrophage:GBM cell co-culture. BMDMs and GBM cells were separated post co-culture using anti-CD45 antibody conjugated magnetic beads. Created in BioRender. Cui, Y. (2025) https://BioRender.com/80×7rf0 B). Heatmap of Log2FC of macrophage MHC-I associated peptide abundance post co-culture compared to unperturbed macrophage (n=3). Fold change was calculated as TMT intensity of co-cultured samples divided by mean TMT intensity of unperturbed samples. C). Correlation plot of mean abundance (n=3) for each peptide in naïve macrophages versus CT2A co-cultured TAMs(left) or GL261 co-cultured TAMs (right). Co-culture upregulated peptides with p<0.05 and log2FC>0.5 were labeled in color. Red lines indicate thresholds for Log2FC=0.5 or 1. D). Reactome pathway enrichment analysis of source proteins from macrophage co-culture upregulated peptides in CT2A and GL261 co-culture. Total proteins profiled from the immunopeptidome was used as enrichment background. P-value was adjusted by Benjamini-Hochberg method. Significant pathways (p<0.05) were highlighted. E). Source proteins associated with cytokine signaling in immune system pathway identified from macrophage co-culture upregulated peptide post CT2A and GL261 co-culture. F). Log2FC of source protein presentation (left) and expression level (right) associated with IFNγ response pathway from macrophages co-cultured with CT2A or GL261.

Corresponding to the increase in MHC-I expression on GAMs compared to naïve macrophages, we observed altered presentation of individual MHC peptides on macrophages polarized by either of the two GBM cell lines, with CT2A-associated macrophage polarization resulting in a greater number of peptides with decreased presentation as compared to GL261-polarized macrophages (Figure 3B, Supplementary Figure 2D). Of the 1309 peptides profiled, 811 had the same directionality of change in presentation during GL261 and CT2A polarization, suggesting that alterations in immunopeptide expression represent a general effect of GBM tumor cells on GAMs. We then focused on peptides with >0.5 Log2FC (>1.41 FC) post co-culture in comparison to the untreated condition (p<0.05) -culture upregulated. Using these criteria, 75 peptides had increased presentation on BMDMs following co-culture with CT2A and 60 peptides had higher presentation following co-culture with GL261 (Figure 3C, Supplementary Table 8). These peptides were presented at varying abundance levels on naïve macrophages. (Figure 3C). Additionally, 53 and 6 peptides decreased in presentation post co-culture with CT2A or GL261, respectively. BMDM showed increased presentation of a similar cohort of source proteins after co-culture with either GL261 or CT2A, with 33 proteins shared between CT2A and GL261 co-culture conditions, accounting for 47% of CT2A induced and 65% of GL261 induced proteins (Supplementary Figure 2E). Two proteins with decreased presentation, Vps33a and Abcb4, were shared between CT2A and GL261 co-cultures.

We performed enrichment analysis on co-culture upregulated source proteins to identify the intracellular response to GBM co-culture that resulted in the observed changes in the immunopeptide repertoires. Reactome pathway enrichment revealed that the increased source proteins were enriched in “Cytokine Signaling in immune system” pathways in both co-culture conditions, suggesting immune function modulation in macrophages through cytokine communication with tumor cells (Figure 3D). Although sharing the same enriched pathways, some contributing source proteins were unique to each co-culture condition. For instance, IL-1b, Stat4, and Gbp7 were restricted to CT2A co-culture while Itgam and Hspa8 were restricted to GL261 co-culture, indicating a differential effect of the GBM cell lines on macrophage peptide presentation (Figure 3E). In addition, we compared co-culture upregulated peptides and source proteins with those upregulated in cytokine-induced pro- or anti-inflammatory macrophages to assess antigen presentation between TAM and canonical macrophages. Only 15 peptides from pro-inflammatory and 2 peptides from anti-inflammatory macrophage states overlapped with co-culture upregulated peptides in either CT2A or GL261 co-culture, demonstrating the distinct antigen repertoire of GAMs from pro- or anti-inflammatory macrophages (Supplementary Figure 2F).

Although magnetic enrichment process resulted in more than 90% purity in macrophage and GBM cell population post co-culture, we wanted to validate that co-culture induced alterations in pMHC expression were not derived from tumor cell impurity in the macrophage population. We thus created MHC-I knock-out GL261 (GL261-KO) and CT2A (CT2A-KO) cells using CRISPR-Cas9 and guide RNAs targeting the signaling sequence and exon1 of both H2-Kb and H2-Db loci. Loss of MHC-I expression in the knock-out cell lines was validated by flow cytometry; label free immunopeptidome analysis was also performed to further confirm that no detectable MHC-I associate peptides were presented (Supplementary Figure 3A). GL261-KO and CT2A-KO tumor cells were co-cultured with BMDM using the same scheme (Figure 2A), and the cell ratio at end of co-culture was similar to that obtained from co-culture with wild-type tumor cells (Supplementary Figure 3B). Samples were directly collected to profile the BMDM immunopeptidome post co-culture because no tumor cell derived MHC-I peptides were present (Supplementary Figure 3C). 73% of the peptides and 77% of source proteins from WT tumor co-culture were also detected in KO-tumor co-culture (Supplementary Figure 3D). TMT intensities were positively correlated between macrophages with KO or WT GBM cell co-culture (R²>0.5), and the correlation is similar to that of the unperturbed macrophage controls between WT and KO experiments, suggesting consistency of the BMDM immunopeptidome using either KO or WT tumor cell lines(Supplementary Figure 3D). GL261-KO co-culture yielded 741 upregulated peptides while CT2A-KO co-culture yielded 124 upregulated peptides on BMDMs (Supplementary Table 9). In the corresponding upregulated proteins, the same Reactome pathways were overrepresented as compared to WT tumor cell co-culture conditions, suggesting that the intracellular responses reflected in the enriched pathways were due to tumor cell induced changes in BMDM rather than impurities during the cell-type separation (Supplementary Figure 3E–F).

We also performed gene set enrichment analysis using source proteins ranked by significance and fold change (-log10 p-value * log2FC) against the Hallmarks gene set from MSigDB. Interestingly, GL261-KO co-culture induced significant positive enrichment of interferon gamma response pathway (Supplementary Figure 3G). Among the interferon gamma response associated proteins, Nfkbia-associated peptides were strongly increased in presentation on BMDMs in both GL261 and CT2A co-cultures (Figure 3F). To understand whether alterations in MHC-I immunopeptide presentation during co-culture are related to coincident changes in protein expression, we performed global protein expression analysis on lysates from BMDM co-culture and naïve BMDM after MHC-I enrichment (Supplementary Table 10). It is worth noting that proteins profiled from this analysis were depleted of MHC-I complexes, potentially along with MHC-I interacting proteins and proteins that are insoluble in MHC-I lysis buffer. Interestingly, we observed differences in the peptide presentation and protein expression profiles for multiple proteins associated with interferon gamma response, including Nfkbia, which showed downregulated protein expression, suggesting that it may be targeted for degradation (Figure 3F, Supplementary Figure 3H). Nfkbia inhibits transcription factor NF-κB, suggesting that its change in expression might modulate NF-κB activity, which regulates macrophage pro- and anti-inflammatory activities(45,46). Indeed, we observed an increased ratio of nuclear:cytosol NF-κB signal in co-culture conditions (Supplementary Figure 4), indicating that decreased protein expression of NFKBia is associated with increased NF-κB transcriptional activity, thereby contributing to maintenance of GAM immunosuppressive phenotype(47). Interestingly, both Stat1 and Stat3 showed increased peptide presentation and protein expression in co-culture with GL261 relative to naïve BMDMs (Figure 3F), despite having opposite regulation on macrophage polarization in response to pro-inflammatory or anti-inflammatory cytokines respectively(48). This result further suggests that co-culture-induced TAMs reflect a macrophage state that is neither pro- nor anti-inflammatory. Together these results suggest that changes in protein expression do not accurately predict changes in MHC-I peptide presentation, in agreement with previous reports(49).

GBM polarized BMDMs present co-culture upregulated peptides at 200–2000 copies per cell, with up to 7-fold increase post co-culture quantified by IS-PRM method

To further explore the co-culture upregulated peptides as potential TAM cell surface targets, we validated the identity of a cohort of co-culture upregulated peptides and performed single point absolute quantification of their presentation on BMDM post co-culture using SureQuant MHC, a targeted MS approach(23,50,51). Peptides were selected for validation and absolute quantification based on the following criteria: profiled in WT and KO GBM cell co-culture conditions with consistent increase in presentation, good spectral match determined by manual validation, predicted to be strong binders of H2-Kb or Db allele, and the source protein is known to be associated with TAM functions (Supplementary Figure 5, Supplementary Table 4). Additionally, co-culture upregulated peptides were selected from Cxcr4 and Tln1, unique to CT2A co-culture and GL261 co-culture respectively. In total, 6 peptides were subjected to SureQuant MHC analysis. This method uses stable heavy-isotope labeled peptides (SIL peptides) to perform triggered acquisition of endogenous peptides, allowing for spectral matching and LC co-elution validation in a single run (Figure 4A). Quantification using the SureQuant method also reduces ratio compression that commonly occurs in TMT-multiplexed measurements. We folded the SIL peptides with empty H2-Kb or H2-Db molecule depending on the peptide binding affinity to generate recombinant heavy-isotope peptide-MHC complexes (hipMHCs). The concentration of each hipMHC was determined by ELISA; each hipMHC was then added to the lysate at 100fmol per sample before MHC peptide isolation workflow (except for Nfkbia and Tln1, which failed to form stable recombinant complex). Addition of the recombinant hipMHC directly to the cell lysate allows for normalization of sample loss during the peptide isolation process, enabling more accurate quantification of the peptide abundance in the sample. All 6 selected peptides were identified in both naïve macrophages and TAMs, with exact matching of MS2 spectra and retention time (Supplementary Figure 6). We approximated the presentation of each peptide on BMDMs by normalizing to the hipMHC (100fmol) and conversion of lysate input to cell number (Figure 4B). Because the two peptides derived from Nfkbia and Tln1 were added in after MHC peptide isolation, the quantification did not account for loss during the isolation process, resulting in underestimated peptide abundance. Supplementary Table 5 shows the measured copies/cell presentation of each of the selected co-culture upregulated peptides in TAMs and naïve macrophages. Among the measured peptides, Csf1r was the most abundant, presented at ~700 copies per cell in naïve BMDM and increased to 1100–1400 copies per cell post co-culture. Cdkn1b and Cxcr4 were presented at ~100 copies per cell in naïve macrophage, and increased to 160–300 copies post co-culture. With the exception of the Siglec1 peptide that did not show differential presentation in SureQuant analysis, all other peptides displayed expected increase in presentation post co-culture with both tumor cells. It is worth mentioning that estimated presentation of Tln1 and Nfkbia peptide was below 100 copies/cell while other peptides were present at hundreds of copies per cell, implying that sample loss during MHC peptide isolation is likely more than 50%, in agreement with other studies(52,53). Overall, GBM cell lines induced increased presentation of selected peptides by 1.2 to 7.6 fold, opening the possibility of the co-culture upregulated peptides to serve as TAM cell surface peptide targets.

Figure 4. Validation and quantification of macrophage co-culture upregulated peptides by SureQuant MHC highlights increased presentation by up to 7.6-fold after co-culture.

A). Schematic of SureQuant workflow. Stable isotope labeled heavy peptide were folded into empty MHC-I complexes and spiked into lysate. MS2 scan of heavy peptide was used to trigger MS2 scan of endogenous light peptide at corresponding m/z offset. Created in BioRender. Cui, Y. (2025) https://BioRender.com/0e2lus6 B). Copies per cell quantification of selected co-culture upregulated peptides in untreated (UT) macrophages and macrophages co-cultured with CT2A/GL261. Significance was calculated using two-tailed student’s t test.

BMDM interactions alter presentation of peptides associated with RhoGTPase signaling pathways in GBM tumor cells

During co-culture with BMDM cells, we observed upregulation of MHC-I expression in both CT2A and GL261 tumor cells (Supplementary Figure 7A). To define the changes in antigen presentation of CT2A and GL261 cells, we profiled the immunopeptide repertoire of tumor cells in mono-culture or co-culture using our quantitative immunopeptidomics platform, as described above. GL261 and CT2A displayed overall similar altered presentation in response to co-culture with BMDM cells, with cohorts of MHC-I peptides increasing (clusters 3 and 4, cluster 2 specific to GL261) and decreasing (cluster 1) in expression (Figure 5A). Although magnetic beads allowed for up to 90% enrichment of tumor cells, residual macrophages can lead to spurious results, as macrophages have much higher levels of MHC presentation than tumor cells. Moreover, peptides from residual macrophages could be represented as significantly increasing in our analysis because these peptides would have been present in co-cultured samples and absent in untreated tumor cell samples (Supplementary Figure 7B). After removal of potential false positive peptides associated with macrophage impurity (Method section: MS data processing for immunopeptidome data), we identified 48 and 108 co-culture upregulated peptides in CT2A and GL261 respectively from 362 profiled peptides (Figure 5B, Supplementary Figure 7C, Supplementary Table 11). 42 of the co-culture upregulated peptides were shared between the two lines, suggesting that the GBM cells also had shared intracellular response to co-culture with macrophages. To help define the processes that contributed to altered antigen presentation, we performed pathway enrichment on the filtered co-culture upregulated source proteins. However, because of limited sample input and therefore limited background size, there were no significantly enriched pathways. Using the whole proteome as background, both total profiled proteins and co-culture upregulated source proteins enriched for Signaling by Rho GTPase related pathways (Supplementary Figure 7D). Rho GTPase pathway is aberrantly regulated in glioblastoma and promotes glioma cell motility and invasion(54–56). The immunopeptidome of other cancer cell lines profiled using label free MS did not enrich for the same pathway, suggesting that presentation of Rho GTPase pathway related proteins might be unique to GBM tumors (Supplementary Figure 7D). Ncf1, Cyfip1, and Ctnnb1 are among the core proteins of these pathways that displayed high fold increase (>2) in presentation, especially in GL261 cells (Figure 5C). We compared the altered peptide presentation of these proteins (Figure 5C) with the changes in source protein expression level during co-culture and found that except for Ctnnb1 in GL261, all proteins displayed slight increase in expression (Supplementary Figure 7E). This increase in expression, however, is not correlated with the high fold change in presentation, supporting previous findings that protein expression is a poor proxy for presentation on MHC-I. To assess the translatability of profiled co-culture upregulated antigens to an in vivo setting, we compared the co-culture upregulated peptides with an immunopeptidome characterization of orthotopic CT2A and GL261 tumors(57). We found 40 out of 48 (83.3%) CT2A-associated co-culture upregulated peptides and 82 out of 108 (75.9%) GL261-associated co-culture upregulated peptides were also presented in vivo. The peptides detected in vivo also included those associated with Rho GTPase related pathways, and are presented at various abundance in vivo (Figure 5D). Additionally, co-culture upregulated peptides from GAMs, including Trem2 and Csf1r, were also detected in vivo, suggesting that these syngeneic tumors had immune cell infiltration (Supplementary Figure 7F). Some peptides were presented at very high abundance in the vivo tumor. For instance, Stat3, a master regulator of cancer hallmarks, is among the top 2% most highly expressed peptides in GL261 and CT2A in vivo tumors, suggesting that they could serve as potential GBM associated antigens in targeted therapy (Figure 5D).

Figure 5. Macrophage interactions alter peptide presentation associated with RhoGTPase signaling pathways in GBM tumor cells.

A). Heatmap of Log2FC of MHC-I associated peptide abundance in CT2A and GL261 post co-culture compared to untreated tumor cells (n=3). B). Overlap of GL261 and CT2A co-culture upregulated peptides and associated source proteins. C). Log2FC of peptides associated with core proteins in RhoGTPase signaling pathways from CT2A and GL261 cells post co-culture. D). Percentile of MHC-I associated peptides from orthotopic CT2A and GL261 tumor ranked by intensity. Overlap with co-culture upregulated peptides from GL261 or CT2A were colored and those associated with RhoGTPase signaling pathway were labeled. E). SureQuant quantification of selected GBM co-culture upregulated peptides from Stat3 and Tlr7. P-value calculated by two tailed student’s t test. F). Schematic of human GBM tumor cell type specific analysis. TAMs (CD45+ population) and GBM cells (CD45− population) were separately analyzed using label free immunopeptidomics approach. Created in BioRender. Cui, Y. (2025) https://BioRender.com/vez2rs5 G). Overlap of peptides identified from human TAMs and GBM population. H). Enriched Reactome pathways of source proteins profiled from TAMs and GBM cells. Whole proteome was used as background. P-value was adjusted by Benjamini-Hochberg method. Pathways are ranked by adjusted p-values and only top 5 are shown.

To validate tumor associated co-culture upregulated peptides, we synthesized SIL peptides of 2 targets, ATL[+7]VFHNL from Stat3 and KGYVF[+10]KEL from Tlr7, that were detected in both GL261 and CT2A and performed SureQuant analysis. ATLVFHNL and KGYVFKEL both demonstrated significantly increased presentation following co-culture with BMDM cells, although to a different level in CT2A and GL261 (Supplementary Figure 8). Corresponding to the higher peptide abundance of Stat3, SureQuant analysis revealed that Stat3 was present at up to 1100 copies per cell post co-culture, while Tlr7 was present at up to 450 copies (Figure 5E, Supplementary Table 6, Supplementary Figure 9). Presentation of the Stat3 peptide increased by over 20-fold in CT2A cells and 3.48-fold in GL261 cells, justifying its potential as a peptide target for GBM immunotherapy.

To determine the translatability of co-culture induced antigen presentation alteration to human GBM tumors, we analyzed the cell type specific immunopeptidome of a patient primary GBM tumor. Cells were extracted from the resected primary GBM tumor and separated into CD45+ and CD45− populations using MACS enrichment (Figure 5F). Tumor cell and TAM immunopeptidomes were separately profiled using label free MS, leading to the identification of 326 peptides from the CD45+ population and 572 peptides from the CD45− population. In line with data from our co-culture studies, CD45+ and CD45− cells have only partially overlapping immunopeptidomes, with 169 peptides identified in both cell types (Figure 5G). Reactome pathway enrichment of profiled source proteins from CD45− population enriched for “RhoGTPase signaling” pathway, while source proteins from CD45+ population were enriched for “Cytokine Signaling in Immune System” (using total proteome as background), similar to the enriched pathways from mouse co-culture upregulated proteins (Figure 5H). This pathway conservation of presented proteins suggests that the in vitro co-culture model of BMDM and GBM tumor cells could capture intracellular processes and antigen presentation patterns that are translatable to the human patient GBM tumor setting. Although peptide identities are different between mouse and human because of source protein sequence differences and MHC binding preferences, we compared the source proteins of the peptides presented in CD45+ and CD45− populations to murine co-culture upregulated source proteins (Supplementary Table 13). In CD45+ cells, we identified 5 homologous proteins from both GL261 and CT2A polarized TAMs, among which PTPN18 and GAPDH were shared (Supplementary Figure 10A). In CD45− cells, we identified 8 and 13 homologues proteins from GL261 and CT2A cells, including 7 proteins that were shared between these two cell lines (Supplementary Figure 10B). The shared homologous proteins span a wide abundance range of profiled human GBM proteins; those with higher abundance could potentially serve as MHC peptide targets in clinical settings. However, it is worth noting that high abundance should not be the only criteria in selecting proper targets for therapeutic applications, and caution needs to be used, especially for peptides presented across multiple tissues. Selecting peptides with high levels of normal tissue presentation could result in an undesired inflammatory response. Nevertheless, this analysis suggests that the cell type specific immunopeptidomics analysis workflow could be applied to identify TAM or tumor cell enriched peptides directly from clinical tissues.

mRNA vaccine against selected co-culture upregulated peptides delays subcutaneous GBM tumor growth

Cancer mRNA vaccines represent a new modality for antigen delivery and enhancement of immune response with promising translation potential in the post pandemic era(58,59). Although co-culture upregulated peptides are self-peptides subject to central tolerance and are presented on multiple cell types, previous studies have shown that self-peptides can generate targeted immune responses in patients(60–62). To evaluate the therapeutic potential of co-culture upregulated peptides, we generated mRNA vaccines encoding 6 selected tumor- or TAM-associated co-culture upregulated peptides that had increased presentation in SureQuant analyses (Supplementary Table 3). Each peptide sequence was separated by an AAY linker (TAA-AAY) or by a furin cleavage site with an ER targeting sequence (TAA-ERISS). In a pilot study, mice bearing GL261 tumors were treated with one dose of TAA-AAY or TAA-ERISS vaccine at Day 7 post tumor cell injection. There was no significant difference in tumor growth with either vaccine, but TAA-AAY resulted in slightly delayed tumor growth post vaccination for 2 weeks (Supplementary Figure 11A). Additionally, 40% of mice had controlled or eradicated tumor post TAA-AAY vaccination (Supplementary Figure 11B). On the other hand, we did not observe any tumor regression or decreased growth rate in TAA-ERISS vaccination, suggesting that choice of cleavage site affects antigen processing and presentation for inducing anti-tumor response. To further improve TAA-AAY vaccine efficacy, we optimized the dosing scheme and added a booster dose of mRNA. Mice with subcutaneous GL261 tumor were vaccinated intramuscularly twice on either Day1 and Day 15 or Day 8 and Day 15 post tumor cell injection and monitored (Figure 6A). Two doses of mRNA vaccine significantly delayed tumor growth, with no significant difference in efficacy between the dosing schemes. (Figure 6B). Although there was no significant difference in survival between mRNA vaccine treated or untreated groups, in both dosing schemes the tumor was eradicated in 1 mouse (14.3% in D1+D15 group, 12.5% D8+D15 group). Additionally, two mice in D1+D15 vaccination group displayed sustained control of tumor before relapse at Day 32 and Day 44 post tumor induction. Interestingly, we observed transient tumor regression after the first dose of mRNA vaccine in mice with an established tumor, but there was no regression in tumor volume in either scheme after the second vaccination, which could suggest antigen escape over time (Supplementary Figure 11C). To assess if the mRNA generated an antigen-specific response, we performed ELISPOT on splenocytes collected from mice 7 days after the second dose. Even though the co-culture upregulated peptides were self-antigens, mice with 2 doses of vaccine generated a readily detectable immune response against the peptide cocktail (Figure 6D, Supplementary Figure 11D). This response was not due to injection of lipid nanoparticles, as mice injected with two doses of SIINFEKL-encoding vaccine at D8 and D15 did not generate a comparable level of response (Supplementary Figure 11E–F). We also evaluated response to each antigen by stimulating splenocytes from vaccinated mice using single peptides. Response towards different vaccinated antigens was highly heterogenous within the vaccinated cohort regardless of dosing scheme (Supplementary Figure 11G). For instance, M3 from D1+D15 and M3 from D8+D15 group showed dominant immune response towards Cxcr4 peptide, while M1 from D1+D15 and M4 from D8+D15 generated similar level of immune response against all vaccinated co-culture upregulated peptides. These data suggest that including a panel of antigens could be beneficial to generate an effective immune response across broader populations. To further explore the impact of co-culture upregulated peptide induced adaptive immune response, we performed immunohistochemistry on subcutaneous tumors 7 days post second dose vaccination. mRNA vaccination resulted in CD8 T cell infiltration in the tumor, suggesting T cell homing upon stimulation with co-culture upregulated antigens (Figure 6F). However, a lower percentage of infiltrated T cells were active in tumors after mRNA vaccination, suggesting exhaustion in the TME (Supplementary Figure 11H). Most importantly, we demonstrated that self-antigen encoding mRNA vaccines effectively generated adaptive T cell response to TAM and GBM antigens in mice bearing GBM tumors, highlighting their therapeutic potential for GBM treatment.

Figure 6. mRNA vaccine against selected co-culture upregulated peptides delayed subcutaneous GBM tumor growth.

A). Schematic of co-culture upregulated peptide encoding mRNA vaccination. Mice bearing subcutaneous GBM tumors were injected with two doses of vaccines at D1+D15 or D7+D15. Created in BioRender. Cui, Y. (2025) https://BioRender.com/w35k12b B). Tumor growth curve comparison of mice with and without mRNA vaccination. Tumor growth was fitted using mixed-effect modeling, and the significance of difference in slopes among groups was determined by two-way ANOVA followed by multiple comparisons. p value was adjusted by Benjamini-Hochberg method. *:p<0.05, **: p<0.01. C). Survival plot of mice with and without mRNA vaccine (n≥6). D). ELISPOT of splenocytes from vaccinated and non-vaccinated mice stimulated with co-culture upregulated peptide antigen cocktail. Representative image was selected from each group. E). Quantification of spots from vaccinated compared to non-vaccinated mice (n=5). Significant difference among groups was determined by Brown-Frosythe and Welch ANOVA tests followed by multiple comparisons. p value was adjusted by Benjamini-Hochberg method. F) Immunohistochemistry of CD8 expression in GBM tumors after mRNA vaccination. Representative images were selected from each group. Scale bar indicates 50μm. The percentage of CD8+ pixels in the tumor area (manually defined to exclude tissue margin) was quantified. Significant differences among groups were determined by ordinary one-way ANOVA tests followed by multiple comparisons (n≥3). p value was adjusted by Benjamini-Hochberg method.

Discussion

The GBM tumor microenvironment is characterized by a high degree of infiltration of blood derived macrophages that contribute to immunosuppression. Conventional chemotherapy and radiotherapy treatment of GBM tumors can increase the abundance and immunosuppressive function of TAMs, resulting in therapeutic resistance and aggressive recurrence(63–66). Targeting TAMs has been shown to sensitize tumors to checkpoint inhibitor treatment and chemotherapy (67,68). However, current TAM depletion strategies for GBM target single survival signaling pathways; these approaches can be limited by incomplete TAM elimination, inefficient drug delivery, and possible compromise of normal macrophage functions(11,69). Defining the antigen presentation dynamics between GBM tumor and macrophages could potentially enable development of targeted immunotherapies against either compartment. This study is the first to quantify cell-type specific changes in antigen presentation during the interaction of GBM tumor cells and macrophages. By comparing TAMs post co-culture to naïve macrophages, we were able to define the temporal change in the immunopeptidome caused by interaction with either of two commonly used murine syngeneic GBM cell lines, thereby connecting altered macrophage states with changes in antigen presentation.

Our results suggest that evolution of antigen presentation in the tumor:immune interface is bi-directional, with GBM tumor cells altering TAM antigen presentation and TAMs altering GBM antigen presentation. Following co-culture, both GBM cell lines presented co-culture upregulated proteins associated with Signaling by Rho GTPase, suggesting a similar response to TAM modulation. However, GL261 had a stronger increase in peptide presentation associated with this pathway, potentially correlated to its higher immunogenicity. We validated that the tumor cell expressed co-culture upregulated peptides were also detected in in vivo orthotopic syngeneic tumors; some of these were among the most abundantly presented peptides in the tumor. Thus, these co-culture upregulated peptides reflecting TAM modulation may serve as targets for GBM cells.

It is worth noting that the accuracy of immunopeptidome analysis can be limited by impurities during MACS separation, especially if the residual cells have much higher MHC-I expression than the enriched cell type. We observed macrophage-derived peptides in the immunopeptide analysis of co-cultured tumor cells and thus we applied post-hoc filtering to remove these residual contaminants. In the analysis of the macrophage immunopeptides, we used MHC null GBM tumor cells to validate that cell enrichment impurities did not affect the altered antigen presentation and biology interpretation.

To test the translational potential of co-culture upregulated peptides, we encoded them in mRNA vaccines and evaluated efficacy in a GL261 subcutaneous model. We observed a significant delay in tumor growth with two doses of mRNA TAA-AAY vaccine at D1+D15 or D8+D15 post tumor injection and tumor eradication in a small portion of mice. Although mRNA vaccine did not significantly improve survival, it is worth pointing out that the analysis may have been confounded by health conditions, as some vaccinated mice had to be euthanized at early timepoints due to ulceration. Interestingly, we only observed transient tumor regression after the first dose of mRNA vaccine in mice with developed tumors. We also observed an immune response against vaccinated co-culture upregulated targets 7 days after the second dose vaccination, suggesting that mRNA encoding self-peptides can break central tolerance and generate an anti-tumor response. Even with generation of antigen-specific T cells, no significant difference in survival was observed, possibly due to T cell exhaustion in the TME(70), suggesting that enhancing T cell activation and sustaining anti-tumor efficacy remains a significant therapeutic challenge of vaccine monotherapy. However, therapeutic combinations using vaccines and immune checkpoint inhibitors or other immune-modulatory strategies could potentially address some treatment challenges and improve clinical outcomes.

We further assessed the potential impact of our cell type specific immunopeptidome workflow by applying it to analyze one patient primary tumor. Although murine in vitro models are known to not recapitulate human biology, it was striking to find that CD45+ and CD45− cells from a human tumor enriched for the same pathways as observed in our co-culture model, suggesting that some cellular responses to TAMs and GBM tumor are conserved across models and organisms.

Overall, this study demonstrates the therapeutic translatability of using cell-type specific MHC peptides to target TAMs and GBM tumor cells in vivo. This approach is generally applicable to a wide range of tumors and can be used in a clinical setting for patient antigen identification to create targeted immunotherapies.

Statement of Significance:

Immunopeptidomic analysis identified altered expression of antigens during macrophage-tumor co-evolution that could be targeted with an mRNA vaccine to significantly inhibit glioblastoma growth, revealing potential immunotherapeutic strategies for treating tumors.