Human study cohort

This study was comprised of three cohorts. First, this study enrolled a local Taiwanese cohort consisting of 28 patients with GBM with complete MRI sets and microarray genomic data obtained from intratumoral and peritumoral tissues through stereotactic sampling. The second cohort was obtained from TCGA database, which included the genomic and survival data of 225 patients with GBM. The third cohort was obtained from CGGA database (mRNA_array_301). All GBM had wild-type isocitrate dehydrogenase 1 (IDH-1). This study was approved by the Institutional Review Board of Taipei Medical University (approval no. N201603086).

Immune subset enrichment level estimation

GBM RNA-seq datasets from TCGA level 1 data were processed and analyzed using HISAT2 or StringTie [69]. The array platform employed for the GBM cohort was Agilent 244 K custom gene expression (G4502A-07–2; Agilent, CA, USA). The gene intensity was normalized using locally weighted scatterplot smoothing. The enrichment levels of various immune subsets, namely TH1, TH2, TH17, cytotoxic T lymphocytes (CTLs), and MDSCs, were evaluated through gene set enrichment analysis based on TCGA Level 1 data. A metagene list is presented in Supplementary Table S1.

Genomic screening of immune-related genes associated with patient survival and survival curve analysis

A total of 8,901 immune-related genes were selected from ImmPort, InnateDB, GO, and the Panther database. Putative survival-associated immune-related genes were ranked and filtered using stringent criteria (hazard ratio [HR] > 1; Wald test, p < 0.01). The association between IL-19 expression level and patient prognosis was assessed using the Cox proportional hazards model, implemented through the coxph function in the R package survival. The HR derived from this model quantified the effect of a one-unit increase in IL-19 expression on the risk of an event occurring. To further examine the impact of IL-19 expression, the cutp function from the R package survMisc was used to identify the optimal cut-off point for stratifying patients into high- and low-expression groups. Using this cut-off value, Kaplan–Meier survival curves were generated to visualize the survival differences between the two groups. The statistical significance of these differences was evaluated using the log-rank test.

RNA microarray and ingenuity pathway analysis of peritumoral and intratumoral GBM tumor tissue

Tissue samples were collected from the intratumoral solid T1 contrast-enhanced tissue and the peritumoral region (tumor-free resection margins of 1–2 cm), which was defined using multimodal magnetic resonance (MR) images (T1, T1 contrast-enhanced tissue (T1 + C), T2 Fluid Attenuated Inversion Recovery, and apparent diffusion coefficient maps); the samples were then obtained using MR-guided stereotaxic surgery. All RNA was extracted from the tumor tissue and hybridized to an Agilent SurePrint Microarray (Agilent Technologies, Design ID: 039494), and the signals were analyzed using feature extraction software (v10.7.3.1; Agilent). An Ingenuity Pathway Analysis (IPA) system (version 42012434, Ingenuity Systems; Qiagen, China) was used for the bioinformatics analysis.

Human tumor tissue immunofluorescence staining

The tumor tissues were fixed with 4% paraformaldehyde for approximately 3 days, then were embedded in paraffin and cut into 5-µm-thick sections, which were deparaffinized and rehydrated using a graded ethanol series. The sections were then exposed to an antigen retrieval process, and sections were stained using IL-19 antibody (R&D systems, #AF1035, 100X dilution), CD206 antibody (Cell signaling, E2L9N, 800X dilution), Iba1 antibody (Cell signaling, E4O4W, 200X dilution). The secondary antibodies were Alexa Fluor 488 AffiniPure donkey anti-rabbit IgG (Jackson ImmunoResearch Laboratories, 500X dilution), and Cy3 donkey anti-goat antibody (Abcam, #ab150, 600X dilution). The sections were observed by a light microscope (Olympus/Bx43).

GBM cell lines and glioblastoma stem cells

Human GBM cell line, including U87, GBM8401, GBM8901, and DBTRG-05MG cell lines were purchased from the Bioresource Collection and Research Center (Taiwan). U118 and LN18 cells were purchased from (American Type Culture Collection, ATCC). Mouse GBM cell line, GL261 cell line, was purchased from the German Collection of Microorganisms and Cell Cultures (Leibniz Institute, DSMZ, Germany). Human glioblastoma stem cells (GSC) were purchased from Celprogen (USA). GL261 cells were transfected with the plasmid vector PGL4.51 (luc2/CMV/Neo). GL261 TMZ-resistant subline (GL261/TMZ-R) cells were established according to our previous study [35]. The DBTRG TMZ-resistant subline (DBTRG/TMZ-R) was generated by long-term incubation with 50 μM TMZ. TMZ-resistance activity was confirmed by a colony formation assay (Supplementary Fig. 1).

Glioma-bearing animal model

C57BL/6JNarl male mice (National Laboratory Animal Center, Taiwan) were used to establish a GBM-bearing animal model. NBSGW mice (NOD, B6.SCID Il2rγ−/− KitW41/W41 mice, Jackson Lab) [45] were used for human GBM-bearing animal model. The tumor cell inoculation assay was performed as previously described in another study [30]. Briefly, 100,000 GL261 or 150,000 GL261/TMZ-R cells were slowly injected into the right brain of C57BL/6JNarl male mice. IL-19 antibody (R&D Systems) or an isotype control antibody was intravenously administered on days 12, 15, 19, and 22 (15 μg/mouse). Tumor volume was analyzed using a Spectrum In Vivo Imaging System (IVIS Lumina III XRMS, PerkinElmer), and the animals were monitored daily to ensure humane endpoints for animal survival experiments. This animal study was approved by the Institutional Animal Care and Use Committee of Taipei Medical University (LAC 2018-0478) and conducted in accordance with the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines and the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health.

Tumor cytokine expression profile

The cytokine profiles of the tumor homogenates were analyzed using the Proteome Profiler Mouse XL Cytokine Array Kit (ARY028; R&D Systems, USA) in accordance with the manufacturer’s instructions. We used 200 μg of the tissue lysates of tumor homogenates to determine cytokine expression profiles. All data were acquired using the ChemiDoc Touch Imaging System (Bio-Rad, USA) and signals were quantified using ImageJ software.

Tumor-infiltrating leukocyte scRNA-seq analysis

ScRNA-seq analysis was performed on day 25 after tumor inoculation to isolate tumor-infiltrating leukocytes from tumor-bearing mice. Mice were anesthetized with zoletil and xylazine and transcardially perfused with cold phosphate-buffered saline (PBS). Tumor tissues were dissociated into single-cell suspensions using a tumor dissociation kit (Miltenyi Biotec) in combination with a gentleMACS Octo Dissociator with Heaters (Miltenyi Biotec). The BD Rhapsody Express Single-Cell Analysis System was used for all the targeted transcriptomic experiments. We first isolated CD45+ live tumor-infiltrating leukocytes and then labeled specific markers (AbSeq Oligo anti-mouse CD3, CD4, CD8, NKp46, CD11b, and CD11c) using a BD Single-Cell Multiplexing Kit (BD Biosciences, #633781, USA) and BD AbSeq Ab-Oligos reagents according to the manufacturer’s protocol. The final libraries were sequenced using the HiSeq 2500 sequencing system (Illumina, USA), which generated paired-end reads of 150 base pairs. The final mean read depths for the two experiments were as follows: Experiment 1 had 8467 reads/cell for the AbSeq library (saturation: 96.5%) and 12,204 reads/cell for the mRNA library (saturation: 67.4%). Experiment 2 had 9950 reads/cell for the AbSeq library (saturation: 97.1%) and 26,728 reads/cell for the mRNA library (saturation: 76.6%). Whole-transcriptome FASTQ files were processed using the standard Rhapsody WTA Analysis Pipeline (version 1.9.1; BD Biosciences) on the Seven Bridges platform, and immune subset clustering was implemented using the Seurat plugin (version 3.8.1) in SEQGEQ software (BD Bioscience). Raw scRNA-seq data were deposited in PRJNA1036784. The DEGs of immune subsets were showed in supplementary file 1.

Generation of Il-19 knockout cell line through CRISPR/Cas9-mediated genome editing

Il-19 knockout (IL-19KO) GL261/TMZ-R cells were generated using CRISPR/Cas9. Two single-guided RNAs (sgRNAs) targeting the second and fourth coding exons of Il-19 were separately cloned into pAll-Cas9. The pPpuro plasmid was obtained from the National RNAi Core Facility (Academia Sinica, Taipei, Taiwan). The sequences of the targeting sites were as follows: 5′- AGAGAATCAACGTCATGCCCAGG -3′ for sgRNA#1, and 5′- GTAGAATGTCAGCAGGTTGTTGG -3′ for sgRNA#2. A control sgRNA targeting EGFP was cloned and used in this study. The two sgRNA plasmids used to generate the Il-19 gene KO GL261/TMZ-R cells were transfected into GL261 cells using the Lonza 4D Nucleofector X Unit (Lonza). Two days post-transfection, the cells were cultured with 2 μg/mL puromycin for 1 week. Viable cells were subjected to limiting dilution in a 96-well plate to isolate single-cell clones. IL-19KO cells were verified by western blot analysis and DNA sequencing of genomic regions.

Generation of Il-19 −/− mice

Il-19−/− mice were generated using CRISPR/Cas9-mediated fragment deletion. Two crRNAs (IDT) targeting introns 1 and 4 were used to generate the Il-19 knockout allele. The crRNA sequences used were 5′-TAGGGATCCTGAAAATGTAG and 5′-GGACTTGGTCCCATGCTAAG, respectively. Super-ovulated 3–4-week-old C57BL/6JNarl female mice were mated with male mice, and one-cell stage zygotes were collected the following day. A mixture of crRNA-tracrRNA and Cas9 recombinant protein (IDT#1,081,061) at concentrations of 0.15 μM, 0.15 μM, and 1.8 μM, respectively, was injected into the cytoplasm of the zygotes. Injected zygotes were transferred into the oviducts of 0.5-dpc pseudo-pregnant ICR female mice (BioLASCO Taiwan). To confirm the knockout allele, three primers (Il19-Fw: 5′-CTACAGTCTTAGGAGATGTCTGATTTCTGT; Il19-Rv: 5′-GTCTACCCAAACATCACACAGC; and Il19 int2-Rv: 5′-CCAAGCCACTGAAATTCTGCCC) were used to differentiate the knockout and wild-type (wt) alleles. The knockout allele and wt allele resulted in amplicons of 278 bp and 531 bp, respectively. Two knockout mouse lines (Il-19−/−-K1 and Il-19−/−-K2) were generated.

DNA constructs and lentiviral transduction

Lentiviral clones expressing a non-overlapping shRNA against human IL-19 (TRCN58725) and LacZ (TRCN0000231722) were obtained from the National RNAi Core Facility (Academia Sinica, Taiwan). 293 T cells were co-transfected with pLKO.1, pCMV-R8.91, and packaging plasmids (pPAX/pMD.2G) using PolyJet (SignaGen Laboratories, USA). After 72 h, the viruses were concentrated by precipitation with the Speedy Lentivirus Purification Kit (Abcam, USA), following the manufacturer’s instructions. For lentiviral transduction, human GSCs were transduced with lentivirus expressing the shRNA and selected with puromycin for 72 h.

Bone marrow derived macrophage supernatant effects on CD8+ T cell activation

Bone marrow-derived macrophages (BMDMs) were differentiated from WT and Il-19−/− bone marrow using M-CSF (100 ng/mL) and then polarized into M2 BMDMs through IL-4 stimulation [38]. CD8+ T cells were isolated from the spleens of C57BL/6JNarl mice using a CD8+ T cell isolation kit (MACS, USA). M2-enriched BMDM supernatants were collected and used to culture 2 × 105 CD8+ T cells in the presence of biotinylated CD3 and CD28 antibodies and anti-biotin MACSiBead particles (MACS, USA) for 3 days. The amount of IFN-γ in the CD8+ T cell culture supernatant was measured using an enzyme-linked immunosorbent assay (ELISA) Max Deluxe Mouse IFN-γ Kit (BioLegend).

Colony formation assay

2000 DBTRG and DBTRG/TMZ-R cells were cultured for 9 days in the presence or absence of 50 μM TMZ. 2000 control (Ctrl) and IL-19KO GL261/TMZ-R cells (KO#1 and KO#2) cells were cultured for 10 days. The cells were then fixed with 4% paraformaldehyde and stained with 0.5% crystal violet. The number of colonies was subsequently counted.

Cell invasion assay

The invasion abilities of GL261/TMZ-R, U118, and DBTRG/TMZ-R cells were analyzed using transwell chambers (Corning Costar 3422, USA) in accordance with the manufacturer’s protocol. A total of 2000 cells were loaded into Matrigel-coated Transwell upper chambers and incubated with IL-19 (100 ng/mL) in the lower wells, which acted as a chemoattractant. The number of cells per field was counted in five random fields of each membrane under an optical microscope. In some experiments, 1 μg/mL WISP1 antibody (R&D Systems) or isotype control antibody was added to the upper and lower chambers.

Western blot

The expression levels of pAkt (Ser473), GAPDH, Phospho-β-Catenin Ser552 (Cell Signaling Technology, USA), IL-20RA and IL-20RB (Thermo, USA), and IL-19 and WISP1 (R&D Systems) were determined in cell lysates by western blotting. Cell lysates were prepared using the PRO-PREP protein extraction solution (iNtRON Biotechnology, Korea) with 2 mM Na3VO4. Horseradish peroxidase-conjugated goat antirabbit, antimouse, or antirat IgG antibody (GoalBio, Taiwan) was used as secondary antibody. All data were acquired using a ChemiDoc Touch Imaging System (Bio-Rad, USA).

T cell activation assay

T cells were isolated from the spleens of C57BL/6JNarl mice using a Pan T Cell Isolation Kit (MACS, USA). TCs were activated using plate-bound CD3 antibody (2C11, 10 μg/mL) and soluble CD28 antibody (37.51, 2 μg/mL) in the presence or absence of IL-19 (50 ng/mL). After 2 days of incubation, the TCs were stimulated with phorbol myristate acetate (PMA, 100 ng/mL) and an ionophore (A23187, 1 μg/mL) for 3 h. Brefeldin A (5 μg/mL) was added during the last 2 h of culturing. The cells were stained with IFN-γ, CD8, and CD4 antibodies in accordance with standard protocols and then analyzed using flow cytometry.

Flow cytometry

Cell surface molecules were stained with specific antibodies in accordance with standard protocols and analyzed using a CytoFlex Flow Cytometer (Beckman Coulter, USA). The following fluorochrome-conjugated antibodies were used to detect IL-19 expression in the tumor-infiltrating cells: IL-19-Alexa Fluor 647 (152112), Ly6G-PB (RB6-8C5), CD45-APC-Cy7 (30-F11), CD11b-PE (M1/70), Ly6C-FITC (HK1.4), CD8-PE (53–6.7), and CD4-PE-Cy7 (GK1.5). The antibodies were purchased from BioLegend or R&D Systems.

Proteomics analysis for GBM cells

First, 1 × 107 GL261 and GL261/TMZ-R cells from each cell line were lysed using 0.5 mL protein extraction buffer (Abcam, USA) with a mass spectrometry-safe protease and phosphatase inhibitor (Sigma, USA). The protein lysates were subsequently reduced and alkylated by adding 2 M urea and 5 mM dithiothreitol (Thermo Scientific, USA) and incubated for 30 min at 37 °C. The mixture was alkylated with 15 mM iodoacetamide (Sigma, USA) for 30 min in the dark at room temperature. The proteins were digested by adding trpsin/LysC (Promega, USA) overnight at room temperature at a 1:50 enzyme-to-protein ratio. The digested samples were acidified with trifluoroacetic acid to achieve a final volumetric concentration of 0.5% and then centrifuged at 15,000 g for 10 min to clear the precipitated urea from the peptide lysates. The samples were dried using a SpeedVac system and desalted using peptide desalting spin columns (Pierce, Thermo Scientific, USA). The peptides were then subjected to reversed-phase fractionation using a high-pH reversed-phase peptide fractionation kit (Pierce, Thermo Scientific, USA) and dried using a SpeedVac system. The dried peptides were desalted using a Ziptip-C18 column (Merck Millipore). Liquid chromatography with tandem mass spectrometry (LC–MS) was performed using a Thermo LTQ Orbitrap Elite mass spectrometer. The peptide mixtures were loaded onto a C18 BEH column 25 cm in length with an inner diameter of 75 μM and packed with 1.7-μM particles with a pore width of 130 Å. The peptides were then separated for 150 min using a segmented gradient from 5 to 35% solvent B (acetonitrile with 0.1% formic acid) at a flow rate of 300 nL/min. The LC–MS raw data were aligned using Progenesis QI software (Waters Corporation), and the proteins were identified using Mascot software.

Procedures of CHOL-PEG-SPIO-IL19 conjugate synthesis

The CHOL-PEG-SPIO-IL19 nanoparticles were constructed in our previous study [34] with some modifications. Briefly, 0.25 mL of amine-functionalized SPIO nanoparticles (1 mg/mL [Fe] = 0.86 nmol/mL nanoparticles, 10 nm in size; Ocean NanoTech, USA) was reacted with 37.5 μL of sulfo-SMCC (10 mg/mL, 858 nmol) at room temperature for 1 h to obtain maleimide-functionalized SPIO nanoparticles. The maleimide-functionalized SPIO nanoparticles were washed with 10 mL of phosphate-buffered saline (PBS) to remove excess free sulfo-SMCC using an LS column (Miltenyi Biotech, Germany) and then eluted in 800 μL of PBS. Subsequently, human IL-19 antibodies (0.5 mg/mL; R&D Systems, USA) were treated with iminothiolane (1.2 μg, 8.7 nmol; Thermo Fisher Scientific, USA), dissolved in 200 μL of Traut’s reagent (50 mM NaHCO3, 150 mM NaCl, and 10 mM EDTA, pH 8.6), and reacted at room temperature for 60 min. After thiolation, 200 μL of 10 mM tris(2-carboxyethyl)phosphine (TCEP; Sigma-Aldrich, USA) was added at room temperature for 30 min. The solution was replaced with 5 mM EDTA in PBS using Vivaspin (10-kDa MWCO polyethersulfone, Sartorius, USA). Finally, the precursor cholesterol poly(ethylene glycol)-thiol (Cholesterol-PEG-SH, MW 2000, 1 mg/mL; NSP, USA), the thiolated antibodies, and the maleimide-functionalized SPIO nanoparticles were mixed and reacted at 4 °C overnight. The unused maleimide-functionalized groups were blocked with excess cysteine for 15 min at room temperature. The CHOL-PEG-modified IL-19 antibody-conjugated SPIO nanoparticles were separated using an MS column and washed with PBS at a volume 25 times greater than the column bed volume to remove unconjugated antibodies. The nanoparticles were then eluted in 600 μL of PBS. The number of immobilized IL-19 antibodies per SPIO nanoparticle was estimated to be two, based on the molarities of the components in the reaction. The total amount of SPIO in the CHOL-PEG-SPIO-IL19 nanoparticles was determined using a spectrophotometric technique (absorption at 500 nm). The fluorescence and absorption signals were measured using the Thermo Varioskan Flash (Thermo Fisher Scientific, USA) and the Multiskan GO Microplate Spectrophotometer (Thermo Fisher Scientific, USA), respectively. CHOL-PEG-SPIO-isotype control nanoparticles were generated using the same procedures, but the IL-19 antibody was replaced with an isotype control antibody (mouse IgG2B, R&D Systems). Unmodified SPIO nanoparticles were used to generate SPIO-IL-19 nanoparticles and followed the same conjugation procedures.

Physicochemical characteristics of CHOL-PEG-SPIO-IL19 conjugate

The measurements of hydrodynamic diameter of the SPIO and SPIO conjugate nanoparticles were determined by Zetasizer Nano ZSP (Malvern Instruments, UK), which using a process called Dynamic Light Scattering (DLS) according to manufacturer instructions. The dispersant of SPIO conjugate nanoparticles was 10 mM NaCl solution. The infrared spectrometer of the SPIO, carriers (Cholesterol-PEG and human IL19 antibody), and SPIO conjugate nanoparticles was determined by subjecting the samples to Fourier-Transform Infrared (FT-IR) Spectroscopy using PERKINELMER FRONTIER (Waltham, USA). The measurement of IR spectrum for the disc of each sample over a wavelength scanning range of 4000 cm−1 to 400 cm−1 to observe the conversion of the functional groups of the copolymer.

To prepare liquid samples containing particles for transmission electron microscopy (TEM), the negative staining technique was employed. Formvar/carbon-coated 200-mesh nickel grids were used as the substrate for sample deposition. A drop of the particle-containing liquid sample was applied onto the grid and allowed to stand for 3 min at room temperature. Excess liquid was carefully removed using filter paper to minimize background interference. Subsequently, the grid was stained with 2% uranyl acetate for 1 min to enhance contrast. After staining, excess uranyl acetate was similarly removed with filter paper. The grid was then air-dried completely before being mounted for TEM observation. TEM images had been acquired by using an HT-7700 scope (Hitachi, Japan), and the analysis was performed according to the manufacturer’s instructions.

To detect the IL-19 expression in human GBM cells, we determined the IL-19 signals through flow cytometry and immunofluorescence assay. The intracellular staining with specific antibodies according to the standard protocols. DBTRG cells were stained with CHOL-PEG-SPIO-IL19 nanoparticles, unconjugated SPIO nanoparticles, IL-19 antibodies (MAB1035, R&D, USA) and isotype control antibodies (402202, BioLegend, USA), followed by FITC-conjugated goat anti-mouse IgG (H + L) antibodies (A-11029, Thermo, USA) by standard protocols and analyzed on a CytoFLEX flow cytometer (Beckman Coulter, USA). Data were analyzed using FlowJo software. The signal of IL-19 expression in DBTRG cells were also determined by immunoflurosence assay with a mounting medium containing 4′6-diamidino-2-dole (DAPI) (Vectashield, USA).

In vitro CHOL-PEG-SPIO-IL19 phantom analysis

MRI images were obtained using a 7 T Bruker PharmaScan MRI scanner having a volume coil with an inner diameter of 72 mm (Bruker BioSpin, MA, USA). T2-weighted images were acquired using spin-echo sequences with an echo time (TE) of 8 ms, a repetition time (TR) of 3000 ms, 50 echoes, a field of view (FOV) of 50 × 50 mm2, a resolution of 256 × 256, and a slice thickness of 1 mm. The MRI samples were CHOL-PEG-SPIO-IL19 nanoparticle phantoms suspended in 1% agarose gel.

Animal MRI measurements

20,000 human GSC cells were injected into the brain of NBSGW mice. In vivo MRI images of mouse brains were obtained using a 7 T Bruker PharmaScan MRI scanner using a volume coil with an inner diameter of 72 mm (Bruker BioSpin). The MRI scanning protocol was followed our previous study [34]. Briefly, MRI was performed in mice anesthetized using 2% isoflurane in the coronal plane. The MRI protocol included a T1 image (TR, 341.3 ms; TE, 4.5 ms; flip angle, 30°; FOV, 16 × 16 mm2; matrix, 256 × 256; 2D; slice thickness, 0.75 mm; number of excitations, 8; resolution, 0.0625 × 0.0625 × 0.75 mm3), a T2-weighted image (TR, 2500 ms; TE, 33 ms; flip angle, 45°; FOV, 16 × 16 mm2; matrix, 256 × 256; 2D; slice thickness, 0.75 mm; number of excitations, 8; resolution, 0.0625 × 0.0625 × 0.75 mm3), a T2*-weighted image (TR, 1000 ms; TE, 12 ms; flip angle; FOV, 16 × 16 mm2; matrix, 256 × 256; 2D; slice thickness, 0.75 mm; number of excitations, 2; resolution, 0.0625 × 0.0625 × 0.75 mm3), SWI (TR, 39 ms; TE, 51.8 ms; flip angle, 15°; FOV, 50 × 50 mm2; matrix, 128 × 128; 3D; slice thickness, 0.5 mm; number of excitations, 3; resolution 0.39 × 0.39 × 0.5 mm3), and T2* mapping (16-echo gradient echo sequence; TR,1150 ms; minimum TE, 3.3 ms; ∆TE, 3 ms; flip angle, 80°; FOV, 16 × 16 mm2; matrix, 256 × 256; 2D; slice thickness, 0.5 mm; number of excitations, 2; resolution, 0.0625 × 0.0625 × 0.5 mm3). The initial T2* mapping scan was performed before injection of CHOL-PEG-SPIO-IL19, SPIO-IL-19, CHOL-PEG-SPIO-isotype ctrl, or SPIO nanoparticles. The second T2* mapping session was started 4 h after injection. The T2* value of each voxel was calculated through exponential fitting performed in-house by using MATLAB (version R2023b, MathWorks, Sherborn, MA, USA). The T2* map was first converted to an R2* relaxivity map by taking its reciprocal. The increase in R2* value (∆R2*) after CHOL-PEG-SPIO-IL19, SPIO-IL-19, CHOL-PEG-SPIO-isotype ctrl, or SPIO injection was calculated using the following formulas: ∆R2* = R2* 4 h − R2* 0 h. R2* values measured in MR images following CHOL-PEG-SPIO-IL19, SPIO-IL-19, CHOL-PEG-SPIO-isotype ctrl, or SPIO administration. The ∆R2* value, which indicates the change in relaxivity due to the local aggregation of CHOL-PEG-SPIO-IL19, SPIO-IL-19, CHOL-PEG-SPIO-isotype ctrl, or SPIO nanoparticles, has a linear relationship with the local CHOL-PEG-SPIO-IL19, SPIO-IL-19, CHOL-PEG-SPIO-isotype ctrl, or SPIO nanoparticle concentration. The coefficient of the linear transformation function between ∆R2* and the local CHOL-PEG-SPIO-IL19, SPIO-IL-19, CHOL-PEG-SPIO-isotype ctrl, or SPIO nanoparticle concentration was estimated through linear regression analysis of agarose phantoms containing various concentrations of CHOL-PEG-SPIO-IL19 nanoparticles. The ∆R2* map was then converted into a local CHOL-PEG-SPIO-IL19, SPIO-IL-19, CHOL-PEG-SPIO-isotype ctrl, or SPIO concentration map by using the estimated linear transformation function. A disk of nickel-coated neodymium iron boron (Nd2Fe14B) with a diameter of 8 mm, a height of 5 mm, and a 0.43-T N42 grade magnet was placed on the tumor site of tumor-bearing mice. After placing the magnet, 50 μg of CHOL-PEG-SPIO-IL19, SPIO-IL-19, CHOL-PEG-SPIO-isotype ctrl, or SPIO nanoparticles was administered via tail vein injection. The magnet was maintained on the tumor site for 1 h and then removed. A radiologist evaluated the MRI images to identify dark signals due to CHOL-PEG-SPIO-IL19, SPIO-IL-19, CHOL-PEG-SPIO-isotype ctrl, or SPIO nanoparticles on T2-weighted images, T2*-weighted images, SWI, and T2* mapping. The tumor volume and hypointense areas in the tumor from T2* and SWI images were calculated using MATLAB. To measure tumor volumes derived from shLacZ and shIL-19 GSCs, T1 MRI images were acquired following administration of the contrast agent Gadovist 1.0 (1 mmol/kg body weight). Post-contrast T1 images were used to delineate tumor boundaries. Tumor margins were manually contoured using the MRIcro software, ensuring precise identification of the enhanced regions corresponding to the tumors. The total tumor volume was calculated by integrating the contoured areas across all relevant image slices using MATLAB, providing a quantitative 3D assessment of tumor size.

Detection of iron in GBM tissues by prussian blue staining assay

To test the binding of CHOL-PEG-SPIO-IL-19 nanoparticles to tumor tissue, CHOL-PEG-SPIO-IL-19 nanoparticles were detected by Abcam iron stain kit (Abcam, USA). After human GSC tumor-bearing mice were injected with CHOL-PEG-SPIO-IL-19 or CHOL-PEG-SPIO-isotype for 4 h, the brains were fixed and embedded in paraffin and cut into 5-μM-thick sections. Brain sections were then deparaffinized, rehydrated through a graded series of ethanol, then directly subjected to a Prussian blue assay to detect the presence of iron caused by CHOL-PEG-SPIO-IL-19 nanoparticles in the tumor tissue. These sections were observed through microscopy (Olympus/BX43).

Statistical analysis

This study conducted Unpaired t-tests and one-way analyses of variance (ANOVA) were used for all data using GraphPad Prism software. Error bars represent the standard deviation from the means.