Glioblastoma (GBM), the most common malignant primary brain cancer in adults, nearly always becomes resistant to current treatments, including the chemotherapeutic temozolomide (TMZ). The long noncoding RNA (lncRNA) TMZ-associated lncRNA in GBM recurrence (lnc-TALC) promotes GBM resistance to TMZ. Exosomes can release biochemical cargo into the tumor microenvironment (TME) or transfer their contents, including lncRNAs, to other cells as a form of intercellular communication. In this study, we found that lnc-TALC could be incorporated into exosomes and transmitted to tumor-associated macrophages (TAM) and could promote M2 polarization of the microglia. This M2 polarization correlated with secretion of the complement components C5/C5a, which occurred downstream of lnc-TALC binding to ENO1 to promote the phosphorylation of p38 MAPK. In addition, C5 promoted the repair of TMZ-induced DNA damage, leading to chemotherapy resistance, and C5a-targeted immunotherapy showed improved efficacy that limited lnc-TALC–mediated TMZ resistance. Our results reveal that exosome-transmitted lnc-TALC could remodel the GBM microenvironment and reduce tumor sensitivity to TMZ chemotherapy, indicating that the lnc-TALC–mediated cross-talk between GBM cells and microglia could attenuate chemotherapy efficacy and pointing to potential combination therapy strategies to overcome TMZ resistance in GBM.
See related Spotlight by Zhao and Xie, p. 1372.
Tumor-associated macrophages (TAM), which can originate from tissue-resident microglia or infiltrating macrophages, are the major biological constituents of the tumor microenvironment (TME) in glioblastoma (GBM; refs. 1, 2). TAMs are implicated in GBM resistance to treatment, angiogenesis, and the colonization and outgrowth of brain metastases (1–3). Exosomes enable communication between GBM cells and TAMs because they allow the exchange of cellular components and activation of signaling pathways in neighboring cells (4). Long noncoding RNAs (lncRNA) can be transmitted via exosomes from cancer cells to other cells in the TME that they modulate, influencing tumor development and/or progression (5). Exploring the roles of intercellular communication via lncRNAs carried in exosomes is important for improving our understanding of the biology of GBM.
LncRNAs, which generally do not encode proteins, can regulate the expression of many normal genes and disease progression (6). In GBM, the lncRNA nuclear enriched abundant transcript 1 (NEAT1) plays a critical role in glioma genesis and tumor progression via an EGFR/NEAT1/EZH2/β-catenin axis (7). Another lncRNA, temozolomide (TMZ)-associated lncRNA in GBM recurrence (lnc-TALC), can regulate the c-Met signaling pathway by competitively binding to miR-20b-3p and activating the Stat3/p300 complex to promote expression of the DNA repair enzyme O6-methylguanine-DNA methyltransferase (MGMT) and resistance to TMZ via modulation of histone H3 acetylation (8). However, whether lnc-TALC exerts these effects by regulating microglia in the GBM microenvironment via exosomes remains poorly understood.
p38 MAPK family members allow cells to interpret a wide range of external signals and respond appropriately by generating a plethora of different biological effects (9). During tumorigenesis in humans and mice, the p38 MAPK signaling pathway is involved in regulating proliferation, differentiation, survival, and migration in a cell context– and cell type–specific manner (10). In GBM, GBM cells release factors that trigger the p38 MAPK pathway in microglia (11). The activity of p38 MAPK is associated with complement components and their receptors, which are involved in the inflammatory status of microglia (12). The complement system plays a critical role in the TME and influences the growth and spread of tumors, thus impacting patient clinical outcomes (13). Therefore, whether the p38 MAPK signaling pathway can change the complement system of microglia, in turn contributing to GBM malignancy, needs to be further explored.
In this study, we found a role for lnc-TALC packaged in GBM cell–derived exosomes in TMZ resistance mediated by reprograming of microglia in the GBM microenvironment and showed that blockade of the lnc-TALC–mediated cross-talk between GBM cells and microglia in the GBM microenvironment can make tumor cells sensitive to TMZ chemotherapy.
Materials and Methods
Transcriptome expression profiling and clinical information data from patients diagnosed with glioma were obtained from The Cancer Genome Atlas (TCGA; https://portal.gdc.cancer.gov) and Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo). The chromatin immunoprecipitation (ChIP) sequencing (ChIP-seq) data were obtained from the SRX100941, SRX100942, SRX150706, SRX186622, SRX186607, SRX186626, SRX100465, SRX3591825, and SRX5846339 data sets in GEO. The binding sites of MEF2C, ZNF770, SREBF1, ZNF518A, ZNF384, TCFL2, CEBPG, ZNF135, ZNF460, and ZNF284 in the promoter region of C5 were predicted on JASPAR data set (http://jaspar.genereg.net).
Cell lines and cell culture
The human glioma cell line LN229, mouse glioma cell line GL261, human microglial cell line HMC3, and mouse microglial cell line BV-2 were purchased from the Chinese Academy of Sciences Cell Bank in 2017. The human astrocyte cell line HA1800, human neuronal cell line HPPNCs, human oligodendrocyte cell line MO3.13 and mouse T follicular helper cells (Tfh cells) were purchased from the Chinese Academy of Sciences Cell Bank in 2021.
We isolated the patient-derived GBM cell line HG7 from discarded GBM tumor tissue in 2016. At that time, the tumor tissue was washed with phosphate-buffered saline (PBS), minced into 1-mm3 pieces, and then digested with 0.1% trypsin (Invitrogen, cat. 27250018) and 10 U/mL of DNase I (Promega, cat. M6101) at 37°C for 45 minutes. Finally, after being washed with ACK lysis buffer (Beyotime, cat. C3702) to lyse the red blood cells, the tissues were triturated by pipetting and passed through a 100-μm cell strainer. These cells were authenticated using the STR assay (Genetic Testing Biotechnology).
Our group previously established TMZ-resistant cell lines named 229R and HG7R from LN229 and HG7 cells, respectively (8).
All cell lines were used at 2 to 6 passages, and negative for Mycoplasma tested using the Mycoplasma Detection Set (ACMEC, cat. CA1080). All the cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM) or DMEM/F12 (Corning, cat. D6429/D6421) supplemented with 10% fetal bovine serum (FBS; BD Biosciences, cat. 04-001-1A) and 1% antibiotic solution (Sigma, cat. A5955) at 37°C in a humidified atmosphere with 5% CO2 and 95% air.
Dimethyl amiloride (DMA, Santa Cruz, cat. 2235-97-4) was added to glioma cells to reduce exosome secretion at a concentration of 25 μg/mL. HMC3 cells were treated with 10 μmol/L AP-III-a4 hydrochloride (ENOblock, MedChemExpress, cat. HY-15858) for 24 hours prior to use in further experiments. SB203580 (Sigma, cat. 152121-47-6) was used to pretreat HMC3 cells at a concentration of 40 μmol/L for 1 hour, and then HMC3 cells were stimulated with exosomes from glioma cells. The p38 MAPK pathway activator P79350 (Calbiochem Inc., cat. 22862-76-6) was dissolved in DMSO, and HMC3 cells were treated with 50 μmol/L P79350 for 1 hour. Aprotinin (Sigma, cat. 9087-70-1) was used to treat microglia at a concentration of 2 mmol/L for 30 minutes. TMZ (Selleck, cat. 85622-93-1) was added to glioma culture medium (CM) at a concentration of 200 μmol/L for 24 hours in vitro and 60 mg/kg via intraperitoneal injection in vivo. Recombinant C5a (R&D Systems, cat. 2150-C5/2037-C5) was used to treat glioma cells at a concentration of 2 μg/mL for 1 hour at room temperature. Different concentrations of C5a receptor antagonist (C5aRa, MedChemExpress, cat. HY-16992A) were administrated for detecting IC50 in vitro as indicated. C5aRa was used in the xenograft model in vivo at a dose of 200 μg/kg/day via intraperitoneal injection. The control group received the same volume of solvent with the same administrative methods.
Establishment of TMZ-resistant cells
TMZ-resistant LN229 and HG7 cells were established as previously described (8). In addition, we seeded GL261 cells into 96-well plates at 6,000 cells per well, and the half-maximal inhibitory concentration (IC50) of TMZ was evaluated. Then, we added TMZ to the cell CM at an IC50 1/50 concentration in 6-well plates. The TMZ dose was increased until GL261 cells grew stably. We kept each dose for 15 days until the end of the fifth month. The induced TMZ-resistant glioma cells were termed LN229R, HG7R, and GL261R.
CCK-8 assay, colony formation assay, and cell apoptosis analysis
The Cell Counting Kit 8 (CCK-8, Dojindo, cat. CK04) was used according to the manufacturer's instructions to evaluate the viability of GBM cells. Viability was measured at OD 450 nm with the BioTek Gen5 system (BioTek).
In colony formation assays, 0.3 × 103 cells per well were seeded in each well of 6-well plates and incubated for 14 days. We washed the resulting colonies twice with PBS and fixed them with 4% formaldehyde for 10 minutes. Then, the colonies were stained for 30 minutes with 0.1% crystal violet (Solarbio, cat. G1061). Olympus camera was used to capture the number of colonies, and ImageJ was used to count them.
For cell apoptosis analysis, either the Annexin V-FITC Apoptosis Detection Kit (BD Biosciences, cat. 556547) or Annexin V-APC Apoptosis Detection Kit (Bender, cat. BMS306) were used according to the manufacturer's instructions. The rates of apoptosis were measured by flow cytometry (FC) using a BD FACSCanto II and analyzed by FlowJo.
RNA isolation, polymerase chain reaction, and quantitative real-time polymerase chain reaction
Total RNA was extracted using TRIzol reagent (Invitrogen, cat. B1277) according to the manufacturer's instructions. The PrimeScript RT reagent Kit (TaKaRa, cat. RR037A) was used to synthesize the cDNAs according to the manufacturer's instructions. In the final reaction volume of 20 μL, 2 μL cDNA were used at a concentration of 100 ng/μL. Real-time quantitative polymerase chain reaction (PCR) was performed on triplicate samples in a reaction mix of SYBR Green (TaKaRa, cat. RR430). Conventional PCR and quantitative real-time PCR (qRT-PCR) were conducted using the CFX96 Touch Real-Time PCR Detection System (Bio-Rad); 1% AGE (agarose gel electrophoresis) was used to detect the PCR products. Indicated genes were normalized to β-actin or U6. The qRT-PCR data were analyzed using the 2–ΔΔCt method. PCR primers for ln-TALC and C5 and the normalization genes were designed using http://www.ncbi.nlm.nih.gov/tools/primer-blast/. Sequences of the primers used in this study are listed in Supplementary Table S1.
Rapid amplification of cDNA ends
Total RNA was isolated with TRIzol reagent (TIANGEN, cat. DP419) according to the manufacturer's instructions. 5′ rapid amplification of cDNA ends (RACE) and 3′RACE analyses were performed with 1 μg of total RNA using the GoScript Reverse Transcription System (Promega, cat. A5000) according to the manufacturer's instructions. RACE PCR products were separated on a 1% agarose gel. The primers used for PCR are listed in Supplementary Table S2.
The fractions of significantly conserved bases and the most conserved 200 nt sliding window (phastCons scores averaged within each window) were used to assess the evolutionary conservation of transcripts. We obtained PhyloP and phastCons scores from the UCSC Genome Browser (14).
CRISPR-Cas9–based lnc-TALC knockdown was performed as previously described (8). Briefly, cells were infected with viruses containing the CAS9 gene and selected with 3 μg/mL puromycin (Selleck, cat. 58-60-6) for 7 days. The cells were subsequently infected with lentivirus carrying small guide RNAs (sgRNA) designed for lnc-TALC. The infection proceeded for 24 hours, and lnc-TALC knockdown was confirmed by qRT-PCR. Details of the sgRNA sequences are shown in Supplementary Table S3.
LN229, HG7, GL261, HMC3, and BV-2 cells were infected with lnc-TALC overexpression lentivirus to generate LN229 OE_lnc, HG7 OE_lnc, GL261 OE_lnc, HMC3 OE_lnc, and BV-2 OE_lnc, respectively. lnc-TALC overexpression lentivirus was synthesized by and purchased from GeneChem Company. To stably silence enolase 1 (ENO1), HMC3 cells were infected with lentiviruses carrying shRNA targeting ENO1 (Santa Cruz Biotechnology, cat. SC-35310) to generate HMC3 OE_lnc shENO1. HMC3 and BV-2 cells were infected with lentiviruses carrying shRNA targeting MEF2C (Santa Cruz Biotechnology, cat. SC-38062) and Mef2c (Santa Cruz Biotechnology, SC-38063), respectively, to generate HMC3 OE_lnc shMEF2C and BV-2 OE_lnc shMef2c.
Exosome isolation and purification
Exosomes were isolated from GBM cell culture supernatants as previously reported (15). In brief, culture supernatants were collected and differentially centrifuged at 300 × g for 10 minutes, 1,000 × g for 20 minutes, and 10,000 × g for 30 minutes. Next, the supernatants were filtered using 0.22-μm filter units (EMD Millipore) and ultracentrifuged at 100,000 × g for 3 hours at 4°C. After removing the supernatant, the pellets were resuspended in ice-cold PBS. Then, the suspensions were centrifuged at 100,000 × g for another 3 hours at 4°C. The exosome-containing pellets were resuspended in PBS and stored at −80°C.
Transmission electron microscopy
Exosome morphology was examined using transmission electron microscopy (TEM). Briefly, exosomes were fixed with 4% paraformaldehyde and placed on 200-mesh Formvar/carbon-coated nickel grids. The grids were fixed with 1% glutaraldehyde followed by washing with distilled water. The samples were exposed to 4% uranyl acetate (UA) for 5 minutes for contrasting. Embedding was done by immersing samples in a 2% methylcellulose solution and 4% UA for 10 minutes on ice. All grids were observed using TEM (JEM-1220EX; Jeol) and the images were recorded with an AMT 2k CCD camera at a primary magnification of ×20,000 to 50,000.
Western blot assay
Total protein was extracted using prechilled RIPA buffer (Solarbio, cat. R0010) combined with 1% proteinase and phosphatase inhibitor cocktails (Selleck, cat. B14001/B15001). According to the manufacturer's instructions, NanoDrop 2000C spectrophotometer (NanoDrop Products) was used to measure the protein to 0.5 to 30 μg/mL via BCA Protein Quantification Kit (Sigma, cat. QPBCA). Each protein sample was subjected to 7.5%/10%/12.5% sodium dodecyl sulfate polyacrylamide gel (EpiZyme Scientific) electrophoresis in a volume of 20 μL. The PVDF membranes were blocked in a 5% milk-TBST solution and then incubated overnight at 4°C with primary antibodies (Supplementary Table S4). All membranes were incubated with HRP-labeled mouse IgG secondary antibodies (Zsbio Store-bio, cat. ZB-2305) and HRP-labeled rabbit IgG secondary antibodies (Zsbio Store-bio, cat. ZB-2306). The chemiluminescence reagent (ECL) kit (Boster, cat. EK1001) was used to visualize the protein bands. Protein quantification was analyzed by Image lab.
Cells were grown on cell coverslips (WHB-24-CS, China) in 24-well tissue culture plates overnight. Then cells were covered with 4% formaldehyde diluted in PBS for 15 minutes in room temperature. After rinsing three times in 1× PBS for 5 minutes each, we treated the cells with 0.5% Triton-X100 (Thermo Fisher, cat. HFH10) and blocked them in blocking buffer (5% bovine serum albumin diluted in warm PBS, BioFroxx, cat. 9048-46-8) for 60 minutes in room temperature. Primary antibodies (Supplementary Table S4) diluted in 1% BSA in PBS were incubated at 4°C overnight. After washing three times with PBS, the cells were incubated with FITC-labeled anti-IgG antibodies (Alexa Fluor 488 and 594, Thermo Fisher, cat. A-110081, cat. A-11011, cat. A-110051, cat. A-110121) for 1 hour at room temperature. DAPI (Sigma, cat. D9542) was used to stain the DNA. Protein subcellular localization was visualized with a fluorescence microscope (Nikon C2).
Enzyme-linked immunosorbent assay
The levels of human TNFα (Abcam, cat. #ab181421), IL1β (Abcam, cat. #ab229384), IL6 (Abcam, cat. #ab46027), TGFβ (Abcam, cat. #ab100647), IL4 (Abcam, cat. #ab215089), and IL10 (Abcam, cat. #ab185986) were measured in the supernatant of human microglia using the noted enzyme-linked immunosorbent assay (ELISA) kits according the manufacturer's instructions.
Fluorescence in situ hybridization
RNA ISH probes were purchased from RiboBio Corporation, and fluorescence in situ hybridization (FISH) was performed using the RNA in situ hybridization kit (BersinBio, cat. Bis-P0001) according to the manufacturer's instructions. In brief, HMC3 and BV-2 cells cultured with exosomes from GBM cells overexpressing lnc-TALC and grown on coverslips overnight were fixed with 4% formaldehyde for 10 minutes, then treated with 0.5% Triton-X100 diluted in PBS for 20 minutes. Before probes were added, the cells were incubated in hybridization mixtures at 37°C for 30 minutes for hybridization. Then the cells were incubated with TALC-FISH Probe Mix at 37°C overnight. After incubation, the cells were washed with 4× saline sodium citrate (containing 0.1% Tween-20) at 42°C for 5 minutes, followed by 2 × SCC at 42°C for 5 minutes and 1× SCC at 42°C for 5 minutes. DAPI (Sigma) was used to counterstain the nuclei, and the high-resolution images were taken using a fluorescence microscope (Nikon C2). The probe sequence is listed in Supplementary Table S5.
Chromatin isolation by RNA purification
Cells were resuspended in precooling PBS buffer (Thermo Fisher, cat. 10010023) and cross-linked with 3% formaldehyde at room temperature on an end-to-end shaker for 30 minutes. The reaction was quenched by adding 125 mmol/L glycine for 5 minutes. The solution was spun at 1,000 RCF for 3 minutes and the supernatant was discarded. One milliliter lysis buffer (50 mmol/L Tris-Cl, 10 mmol/L EDTA, 1% SDS) was added to each 2 × 107 cells, and the samples were sonicated in an ice-water bath until the cell lysate was no longer turbid. After being spun at top speed, supernatant was transferred to 2 volumes of hybridization buffer (750 mmol/L NaCl, 1% SDS, 50 mmol/L Tris-Cl, 1 mmol/L EDTA, 15% formamide), mixed well and incubated at 37°C overnight. Prebind biotinylated oligo probe (4 probes for test group, 1 probe for negative control and positive U1 control, 100 pmol per 2 × 107 cells) was added to streptavidin beads (Thermo Fisher, cat. 65002) for 30 minutes and then mixed with the cell lysate and hybridized at 37°C overnight on an end-to-end shaker. The beads were washed five times with 1 mL prewarming wash buffer (2 × NaCl and sodium citrate, 0.5% SDS) for 5 minutes. At the last washing, 1/20 beads were transferred for qPCR analysis. Elution buffer (20 mmol/L Tris-HCl, 0.05% SDS, 0.5 mmol/L TCEP, 2 mmol/L MgCl2) in a volume of 100 μL and 20 U Benzonase was used to elute protein at 37°C for 1 hour. The supernatant was transferred to a new low binding Eppendorf tube to elute the beads again with 100 μL elution buffer. After combining the two supernatants, we reversed the crosslink at 95°C for 30 minutes. The protein was precipitated with 0.1% sodium deoxycholate (SDC, Sigma, cat. D5670) and 10% trichloroacetic acid (TCA; Sigma, cat. T0699) at 4°C for 2 hours. After the samples were spun at top speed, the pellets were washed with pre-cold 80% acetone 3 times. The probe sequences of lnc-TALC are listed in Supplementary Table S6.
Liquid chromatography-tandem mass spectrometry analysis
Half of the peptides in each sample derived from HMC3 cells overexpressing lnc-TALC were separated and analyzed with a nano-UPLC (EASY-nLC1200) coupled to Q-Exactive mass spectrometry (Thermo Finnigan). A reverse-phase column (100 μm, ID × 15 cm, Reprosil-Pur 120 C18-AQ, 1.9 μm) was used to perform the separation. H2O with 0.1% formic acid (FA) and 2% acetonitrile (ACN; phase A) or 80% ACN and 0.1% FA (phase B) formed the mobile phases. We executed a 120-minute gradient at a 300 nL/min flow rate to separate each sample. Gradient B consisted of 8% to 30% for 92 minutes, 30% to 40% for 20 minutes, 40% to 100% for 2 minutes, 100% for 2 minutes, 100% to 2% for 2 minutes, and 2% for 2 minutes. An Orbitrap analyzer was used for a data-dependent acquisition in the profile and positive mode at a resolution of 70,000 (200 m/z) and m/z range of 350 to 1,600 for MS1, and the resolution was set to 17,500 with a dynamic first mass for MS2. The automatic gain control (AGC) target for MS1 was set to 1 × 106 with a max IT of 100 ms, and 5 × 104 for MS2 with a max IT of 200 ms. The top 10 most intense ions were fragmented by higher energy collisional dissociation with normalized collision energy of 27% and an isolation window of 2 m/z. The dynamic exclusion time window was 30 seconds.
ENO1 activity assay
NADH oxidation was performed to measure ENO1 activity in a pyruvate kinase–lactate dehydrogenase coupled assay as previously described (16). In brief, 20 mmol/L Tris-HCl, 1 mmol/L EDTA, and 1 mmol/L β-mercaptoethanol (pH 7.4) were used to lyse cells. A polytron homogenizer was used to homogenize the lysate three times for a period of 10 seconds followed by sonication. This assay was based on the formation of ATP linked to the disappearance of NADH via pyruvate kinase and lactate dehydrogenase. We measured oxidation of NADH either spectrophotometrically by absorbance at 340 nm or fluorescently by excitation at 340 nm and emission at 460 nm to record enolase activity with NanoDrop 2000C spectrophotometer (NanoDrop Products). Enzyme activities were expressed as units per milligram of protein (1 unit = 1 μmol of substrate converted per minute).
Luciferase reporter assay
We seeded HMC3 cells at 4 × 104 cells/well in 24-well plates and allowed them to settle overnight. The next day, cells were transfected with GV272 luciferase vector (GeneChem) in 12.5 μL DMEM with 10% FBS and 0.2 μL Lipofectamine 2000 (Thermo Fisher, cat. 11668030) for 48 hours according to the manufacturer's instructions. C5 wild-type with potential MEF2C binding sites or mutants of each binding site were amplified by PCR and cloned into the XbaI site of the firefly dual-luciferase expression vector linearized by 50 μL restriction enzyme reaction system. Then, cell lysates were prepared and quantified using the Dual-Luciferase Reporter Assay System (Promega, cat. E2920). Both firefly and Renilla luciferase activities were calculated, and the firefly/Renilla luciferase activity was recorded as fold induction.
RNA immunoprecipitation (RIP) was performed with a Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, cat. 17-700) according to the manufacturer's instructions. In brief, the cell lysates of HMC3 cells overexpressing lnc-TALC with RNase inhibitor and protease inhibitor were centrifuged. Then, the cell lysates were incubated with magnetic beads coats with the indicated antibodies at 4°C overnight. Proteinase K was used to treat the bead-bound immunocomplexes at 55°C for 30 minutes, then the samples were washed with RIP wash buffer. Samples were then centrifuged at 14,000 rpm for 15 minutes and placed on a magnetic separator for isolating RNAs. The RNA fraction precipitated by RIP was analyzed via qPCR. The RIP-PCR products were detected with 4.8% agarose gel electrophoresis. The antibody (Cell Signaling Technology, cat. #3810) used in the RIP assays of ENO1 was purchased from Cell Signaling Technology.
The Millipore EZ-Magna ChIP kit (cat. #17-371) was used for all ChIP experiments. In brief, 2 × 105 cells were crosslinked with 1% formaldehyde for 10 minutes at room temperature. Then, 10 × glycine was used to quench the cross-linking. We sonicated chromatin to 200 to 1,000 bp in lysis buffer, and then extraction of ChIP DNA was performed following the kit's protocol. The MEF2C-specific antibody used is listed in Supplementary Table S4 (Abcam, cat. ab211493).
For the comet assay, cells with different treatments were collected and resuspended in 100 μL of 10% low-melting-point agarose (Promega, cat. V2111). Then, the monosuspension was cast on a microscope slide and gelled at 4°C for 20 minutes. The slides were immersed in the cell lysis solution (Trevigen, cat. 7500-100-06), and the cells were lysed at 4°C for 1 to 2 hours. Then the slides were washed in distilled water and immersed in electrophoresis solution (Y-J Biological, cat. YS-10169RJ). An electric field was applied (25 V, 300 mA) for 30 minutes. The slides were then neutralized, stained with 1× SYBR Green (Sangon, cat. 163795-75-3), and analyzed with a microscope. Tail length and percentage of DNA in tail indicating the level of DNA damage were analyzed by ImageJ in 100 cells of each slide.
Hematoxylin–eosin staining and IHC
The mouse GBM tissues derived from GL261 scramble cells, GL261 cells overexpressing lnc-TALC or 261R cells were fixed in 4% paraformaldehyde for 24 to 48 hours and then dehydrated and embedded in paraffin using standard procedures. For hematoxylin–eosin (H&E) staining, after deparaffinization and rehydration, the slides were stained in hematoxylin solution (ZSGB-BIO, cat. ZLI-9610) for 8 minutes and in eosin–phloxine solution (ZSGB-BIO, cat. ZLI-9613) for 30 seconds to 1 minute, then mounted with xylene. For IHC, 4-μm paraffin sections were stained with a DAB staining kit (ZSGB-BIO, cat. ZIL-9017). Briefly, formalin-fixed, paraffin-embedded tissue sections were incubated at 80°C for 15 minutes, dewaxed in xylene, rinsed in graded ethanol, and rehydrated in double-distilled water. To restore antigen, the slides were pretreated by steaming in sodium citrate buffer for 15 minutes at 95°C. Then the slices were incubated with indicated primary antibodies (Supplementary Table S4) at 4°C overnight. After incubation, the slices were incubated with HRP-labeled either anti-mouse IgG or anti-rabbit IgG (ZSGB-BIO, cat. DS-0003/DS-0004) secondary for 30 minutes. The slices were stained with DAB chromogen solution and incubated for 2 minutes, rinsed in PBS, and counterstained with hematoxylin.
Quantitative evaluation was performed by examining each section using at least 10 different high-power fields with the most abundant stained cells (17, 18). Two independent neuropathologists reviewed and scored each stained slide in a double blinded fashion. The proportion of stained cells counts per field was used for statistical analysis. The intensity was graded depending on the number of positive cells seen: “−” no staining, “+” weak staining, “++” moderate staining, and “+++” intense staining. The representative imaged fields were determined by the average method.
Xenograft model in vivo
C57BL/6 mice were purchased from the animal center of Beijing Vital River Laboratory Animal Technology and bred under special pathogen-free conditions. All animal experiments followed protocols approved by the Institutional Committee on Animal Care of the Second Affiliated Hospital of Harbin Medical University. The total number of 0.5–1 × 106 GL261 cells overexpressing lnc-TALC or the scramble cells transfected with luciferase lentivirus were stereotactically injected into the brain, at coordinate 1 mm anterior and 2 mm lateral of the right hemisphere relative to the bregma, at a depth of 4 mm. After tumors had grown for 7 days, mice were separated into three groups that were treated with control, TMZ alone, or C5aRa combined with TMZ for 7 days at the doses indicated above (see Drug treatment). After 0, 10, and 20 days, the intracranial tumors were measured with bioluminescence imaging via intraperitoneal injecting D-Luciferin, Potassium Salt (Yeason, cat. 4902ES) at a concentration of 150 mg/kg using an IVIS Lumina Imaging System (Xenogen). Mice bearing GBM that exhibited obvious weight loss (≥20% before the experiment) or onset of significant neurologic symptoms, such as seizures, impaired balance, and hemiplegia were considered in a moribund condition. Mice were euthanized via CO2 asphyxiation (10%–20%/min) when these symptoms were identified. After all the mice were dead, the tissues of the mice were carefully extracted and fixed in 10% formalin for the H&E and IHC staining, and blood samples were collected and mixed with potassium ethylenediaminetetraacetic acid (EDTA) at a concentration of 1.5 mg/mL for routine blood and serum enzyme assays by Hematology Analyzer XE2100 (SYSMEX) and Fully Automatic Biochemistry Analyzer Dimension (SIEMENS) at the Department of Laboratory Diagnosis of The Second Affiliated Hospital of Harbin Medical University.
Ethics approval and consent to participate
We obtained human glioma tissues from glioma patients being treated at the Second Affiliated Hospital of Harbin Medical University. In this study, fresh GBM tissues were collected from 10 patients by surgical resection (Department of Neurosurgery, The Second Affiliated Hospital of Harbin Medical University) from 2018 to 2019. The inclusion criteria were as follows: patients treated by surgical resection and pathologic sampling; patients who were willing to provide disease-related materials, including pathologic diagnosis, relevant document records, medical records, and discarded tumor tissues from surgical resection; patients who were willing to provide contact information and maintain follow-up; patients who did not participate in other clinical studies; patients who were willing to write informed consent of the patients and/or the legal guardians; the researcher believed that participating in this study would not affect the treatment effect. Patients who did not meet any of the above criteria were excluded from this study. Tumor tissues were stored by quick freezing until isolating the patient-derived GBM cell line. Informed consent was obtained from patients enrolled in this study and the Clinical Research Ethics Committee of the Second Affiliated Hospital of Harbin Medical University approved the study protocol.
Significant differences between the groups were estimated by the Student t test. Overall survival curves were used to describe survival distributions, and the log-rank test was applied to assess statistical significance between different groups. The survival data were further processed by using univariate and multivariate Cox regression analyses. Pearson correlation coefficients were used to analyze the correlations between variables. Comparisons between quantitative evaluation of IHC assays were performed via the Chi-squared test. Gene ontology (GO) pathway analysis was performed using the DAVID website (https://david.ncifcrf.gov) based on the GSE139549 data set from GEO. R version 4.0.3 with the extension packages pheatmap, ggplot2, and limma was used to produce figures. All results are expressed as the mean ± SD. All statistical analyses were performed using GraphPad software version 7.0 (GraphPad Software) or IBM SPSS Statistics 23.0 (SPSS). Bowtie2, Samtools, deeptools, and MACS2 were used for ChIP-seq analysis. Integrative Genomics Viewer was used for the visualization of ChIP-seq data. A value of P < 0.05 was considered statistically significant.
Exosomal lnc-TALC derived from GBM cells promotes M2 polarization of microglia
To identify mechanisms of resistance to TMZ, we established GL261 cells resistant to the chemotherapeutic, referred to as 261R cells. Compared with the GL261 parental cells, 261R cells showed a poor response to TMZ, as illustrated by an increased half-maximal inhibitory concentration (IC50), enhanced independent growth ability, and decreased apoptosis under TMZ treatment (Supplementary Fig. S1A–S1C). Utilizing the BLAST-like alignment tool, we identified predicted regions of lnc-TALC in the human genome and found that lnc-TALC had similar identity and elevated scores compared with NEAT1 and TP53 (Supplementary Fig. S1D). Phylogenetic analysis using the UCSC Genome Browser (http://genome.ucsc.edu/) revealed that lnc-TALC was highly conserved among mammals and vertebrates, suggesting an important function for lnc-TALC (Supplementary Fig. S1E). However, RACE and qRT-PCR assays showed that lnc-TALC was absent in mouse Tfh cells and GL261 cells (Supplementary Fig. S1F–S1H).
Oncogenic lncRNA-enriched exosomes can remodel the TME to facilitate tumor growth and development (19). To investigate the biological functions of exosomes in the GBM microenvironment, we first purified exosomes from GBM-derived conditioned medium (GCM) collected from TMZ-resistant GBM cell and parental cell cultures and then confirmed exosome identity by TEM (Supplementary Fig. S2A). Particle counting indicated that there was a modest increase of exosome secretion in TMZ-resistant GBM cells compared with the parental cells (Supplementary Fig. S2B). We performed a qRT-PCR assay to compare the expression level of lnc-TALC between the exosomes from TMZ-resistant GBM cells and those from the parental cells and found that those from the TMZ-resistant GBM cells had a higher lnc-TALC expression level (Fig. 1A). In addition, we performed qRT-PCR assays to detect the expression level of lnc-TALC in exosomes from a human astrocyte cell line HA1800, human neuronal cell line HPPNCs, and the human oligodendrocyte cell line MO3.13 and found that lnc-TALC was absent in the exosomes produced by these cell types (Supplementary Fig. S2C–S2E). The levels of lnc-TALC in GCM were unchanged upon RNase treatment but significantly decreased upon simultaneous treatment with RNase and Triton X-100, indicating that the lnc-TALC was mainly enclosed by a membrane, rather than being directly released into the medium (Fig. 1B). Furthermore, we stably overexpressed lnc-TALC (OE_lnc) in the parental glioma cell lines LN229, HG7, and GL261 (Fig. 1C and D; Supplementary Fig. S2F). TSG101 and PTRF (20, 21), which can regulate exosome biogenesis, were determined to be higher in the cells overexpressing lnc-TALC than in the scramble cells (OE_scra) via Western blot (WB; Supplementary Fig. S2G). An immunofluorescence (IF) assay captured Dil-labeled exosomes internalized by the human microglial cell line HMC3 cells during a 2-hour incubation (Supplementary Fig. S2H). These results indicated that lnc-TALC expression levels might be associated with more secreted exosomes in GBM cells overexpressing lnc-TALC than in the parental cells. In GBM, TAMs mainly originate from brain-resident microglia and can create a supportive stroma for neoplastic cell expansion and invasion (22). A qRT-PCR assay showed that exosomes from LN229 glioma cells overexpressing lnc-TALC had higher levels of lnc-TALC than those from scramble cells (Fig. 1E).
To determine whether exosome-transmitted lnc-TALC in GCM was able to reprogram microglia, we cultured HMC3 microglial cells with the exosomes from LN229 glioma cells overexpressing lnc-TALC or scramble cells (Fig. 1F). After treatment with exosomes from the cells overexpressing lnc-TALC, HMC3 cells had a significantly increased level of lnc-TALC (Fig. 1G). We performed WB and IF assays to detect the expression of M2 markers (Arg-1 and CD163) and M1 markers (TNFα and iNOS) to determine the polarization phenotype of the microglia. HMC3 cells treated with exosomes from GBM cells overexpressing lnc-TALC had a marked increase in expression of M2 markers (Fig. 1H–J; Supplementary Fig. S2I). M2-related cytokines were upregulated in the supernatants of HMC3 cells treated with exosomes from LN229 glioma cells overexpressing lnc-TALC, as shown by ELISA (Fig. 1K). To further determine whether lnc-TALC promotes M2 polarization of microglia, GL261 cells overexpressing lnc-TALC and the scramble cells were injected into the brains of C57BL/6 mice. IHC showed that expression of the M2 markers Arg-1 and Cd163 was increased in mice bearing tumors from GL261 cells overexpressing lnc-TALC compared with those from the scramble cells (Fig. 1L and M). HMC3 microglial cells cultured with exosomes from DMSO-treated LN229 glioma cells overexpressing lnc-TALC had higher levels of expression of M2 markers and M2-related cytokines (TGFβ, IL4, and IL10) than HMC3 cells cultured with exosomes from the same cells treated with DMA, an exosome-release inhibitor (Supplementary Fig. S3A–S3D).
We next stably decreased the expression levels of lnc-TALC in TMZ-resistant cells and their exosomes by knocking down lnc-TALC in TMZ-resistant 229R cells with a CRISPR-Cas9 system (Supplementary Fig. S3E and S3F). HMC3 cells cultured with exosomes from TMZ-resistant 229R cells with lnc-TALC knockdown had lower expression levels of lnc-TALC and M2 markers than those cultured with exosomes from TMZ-resistant scramble 229R cells (Supplementary Fig. S3G–S3J). Consistent with these data, ELISAs showed that the levels of M2-related cytokines in the supernatants of HMC3 microglial cells treated with exosomes from TMZ-resistant 229R cells with lnc-TALC knockdown were decreased (Supplementary Fig. S3K). Overall, these data demonstrate that the secretion of exosomal lnc-TALC by GBM cells can result in M2 polarization of microglia in vitro.
lnc-TALC binds with ENO1 to promote phosphorylation of p38 in microglia
LncRNAs can accomplish many of their diverse functions through direct interactions with RNA-binding proteins (23). To determine the potential mechanism by which lnc-TALC induces M2 polarization of microglia, we first overexpressed lnc-TALC in HMC3 microglial cell cells and validated its overexpression through qRT-PCR analysis (Supplementary Fig. S4A). We then sought to identify intracellular lnc-TALC–binding factors using liquid chromatography-tandem mass spectrometry (LC-MS/MS) after lnc-TALC was captured from total cellular extracts of these cells through chromatin isolation by RNA purification (ChIRP; Fig. 2A; Supplementary Fig. S4B and S4C). By comprehensively analyzing a number of proteins, unique peptides, and scores, we identified ENO1 as a potential lnc-TALC–binding protein for further validation (Fig. 2B; Supplementary Table S7). We confirmed that ENO1 interacted with lnc-TALC by WB (Fig. 2C). Moreover, an RIP assay showed that lnc-TALC interacted with ENO1 in HMC3 cells overexpressing lnc-TALC (Fig. 2D). HMC3 and BV-2 cells were cultured with exosomes from LN229 and GL261 cells overexpressing lnc-TALC, and we found that lnc-TALC and ENO1 were enriched in the cytoplasm (Supplementary Fig. S5A–S5C). T-Coffee Multiple Sequence Alignment (https://www.ebi.ac.uk/Tools/msa/tcoffee/) analysis revealed that ENO1, the major enolase isoform in GBM cells (16), was conserved across mammalian evolution, with the whole region of the protein being highly conserved (Supplementary Fig. S5D). We also stably overexpressed lnc-TALC in mouse microglial BV-2 cells (Supplementary Fig. S5E). The relative activity of ENO1 was enhanced in both HMC3 and BV-2 cells overexpressing lnc-TALC (Fig. 2E and F).
To further elucidate the functional role of ENO1 in microglia, we performed GO analysis based on the GSE139549 data set and observed that ENO1 was correlated with the MAPK signaling pathway (Supplementary Fig. S5F; Supplementary Table S8). Inhibition of ENO1 with ENOblock decreased the phosphorylation of AMPK and p38 in HMC3 cells overexpressing lnc-TALC (Supplementary Fig. S5G). In addition, knockdown of ENO1 inhibited the phosphorylation of AMPK and p38 in HMC3 cells overexpressing lnc-TALC (Supplementary Fig. S5H). Taken together, these results indicated that lnc-TALC could promote the phosphorylation of p38 by binding ENO1 in microglia.
Exosomal lnc-TALC elevates C5 expression
Based on the above results, we treated microglia with exosomes derived from LN229 cells overexpressing lnc-TALC or scramble and performed transcriptional profiling. This revealed that microglia exposed to exosomes from GBM cells overexpressing lnc-TALC had higher levels of expression of C5 than those treated with exosomes from scramble cells (Fig. 3A). In our previous study, we found that C5, as a DNA damage repair (DDR)–related cytokine, might be associated with M2 polarization of microglia involved in remodeling the immunosuppressive GBM microenvironment (24).
Next, we investigated whether the p38 MAPK signaling activated by lnc-TALC from LN229 exosomes was related to the expression level of C5. WB and IF assays revealed that expression of C5 was higher in HMC3 cells cultured with exosomes from LN229 cells overexpressing lnc-TALC than in those cultured with exosomes from scramble cells (Fig. 3B; Supplementary Fig. S6A and S6B). Moreover, expression of C5 was decreased in HMC3 and BV-2 cells after culture with TMZ-resistant 229R cells in which lnc-TALC was knocked down. In addition, the levels of expression of Arg-1 and CD163 were different, following the same pattern as the alterations in C5 expression (Fig. 3B; Supplementary Fig. S6A and S6B).
Urokinase (uPA)+ macrophages can regulate C5a release in a C3-independent manner during premalignant progression (25). Aprotinin, a competitive serine protease inhibitor, reversed the elevated C5a level in HMC3 and BV-2 microglial cells cultured with exosomes from LN229 and GL261 cells overexpressing lnc-TALC, respectively (Supplementary Fig. S6C). Consistent with our previous results, administration of the p38 inhibitor SB203580 decreased the level of expression of C5 in HMC3 cells cultured with exosomes from LN229 cells overexpressing lnc-TALC, and the p38 activator P79350 had a positive regulatory effect on C5 expression in HMC3 cells cultured with exosomes from scramble cells (Fig. 3C; Supplementary Fig. S6D). PCR results showed that SB203580 significantly decreased levels of expression of C5 mRNA in HMC3 cells treated with exosomes from LN229 cells overexpressing lnc-TALC (Fig. 3D). Conversely, P79350 increased C5 mRNA expression in HMC3 cells treated with exosomes from scramble cells (Fig. 3D). These results suggested that lnc-TALC regulated C5 expression at the transcriptional level by activating p38.
Then, we predicted the transcription factors potentially binding to the C5 promoter region in the JASPAR data set (http://jaspar.genereg.net; Fig. 3E; Supplementary Table S9). To elucidate the mechanisms by which transcription factors regulate C5 expression, we used TCF7L2 (SRX100941 and SRX100942), SREBF1 (SRX150706 and SRX186622), ZNF384 (SRX186607 and SRX186626), and MEF2C (SRX100465 and SRX3591825) ChIP-seq data to analyze the binding sites of these transcription factors in the promoter region of the C5 gene (Fig. 3F; Supplementary Fig. S7A and S7B), and found that there was an enrichment peak for MEF2C in the promoter region of C5. Potential binding sites for MEF2C were found in the promoter region of the C5 gene (Fig. 3G). In addition, we analyzed ChIP-seq data (SRX5846339) to predict the binding sites of Mef2c in the promoter regions of the C5 gene in mice (Supplementary Fig. S7C). A ChIP-PCR assay detected increased enrichment of MEF2C binding sites in the promoter region of C5 in HMC3 cells treated with exosomes from LN229 cells overexpressing lnc-TALC compared with those treated with exosomes from scramble cells (Fig. 3H). p38 MAPK interacts with MEF2C and enhances the transcriptional activity of MEF2C in inflammation (26). Luciferase activity assays showed that mutation of binding sites 2 and 3 abolished reporter activity, whereas mutation of binding site 1 only reduced the signal by approximately 10% (Fig. 3I). These data suggested that binding sites 2 and 3 had the main roles in mediating C5 binding. While further verifying the effect of MEF2C on C5 expression, we found that knockdown of MEF2C decreased expression of C5 in both HMC3 and BV-2 cells overexpressing lnc-TALC (Fig. 3J; Supplementary Fig. S7D). Collectively, these findings demonstrated that lnc-TALC transferred by exosomes from GBM cells could increase C5a release in microglia via the p38 MAPK signaling pathway.
Microglia-derived C5a induced by lnc-TALC promotes GBM resistance to TMZ
We used expression profiling of 143 DDR genes (27) in TCGA to delineate how C5 variation associated with alterations in the DDR process and identified that mismatch excision repair (MMR), base excision repair, nucleotide excision repair, nonhomologous end-joining (NHEJ), homologous recombination, and Fanconi anemia all were positively correlated with C5 expression (Fig. 4A; Supplementary Fig. S8). To further verify that C5a had effect on DDR in GBM cells, LN229 cells were first treated with recombinant C5a. After subsequent TMZ treatment for 24 hours, there were higher levels of pATM, pATR, and RAD51 and lower levels of γH2AX in the LN229 cells treated with recombinant C5a than in those treated with DMSO (Fig. 4B; Supplementary Fig. S9A). In addition, we found that recombinant C5a treatment caused less TMZ-induced DNA damage than DMSO treatment in LN229 cells in comet assays (Fig. 4C). FC and IF assays were performed to validate that recombinant C5a attenuated TMZ-induced DNA damage and apoptosis in GBM cells cultured with culture medium (CM) from HMC3 cells overexpressing lnc-TALC (Fig. 4D–F).
LN229 cells cultured with CM from HMC3 cells overexpressing lnc-TALC had higher levels of expression of pATM, pATR, and RAD51 and lower levels of expression of γH2AX than those in the control group after 24 hours of TMZ treatment (Fig. 5A; Supplementary Fig. S9B). LN229 and HG7 cells cultured with CM from HMC3 cells overexpressing lnc-TALC exhibited reduced DNA damage when exposed to TMZ compared with those cultured with CM from cells in the control group (Fig. 5B). To explore whether blocking the C5a receptor could improve the sensitivity of GBM cells to TMZ chemotherapy, C5aRa was used to treat LN229 and HG7 cells cultured with CM from HMC3 cells overexpressing lnc-TALC. We elucidated the appropriate concentration of C5aRa to administer to LN229 and HG7 cells by measuring the IC50 and chose 3 nmol/L for further experiments (Fig. 5C). Compared with DMSO, C5aRa decreased the levels of pATM, pATR, and RAD51 and increased the level of γH2AX in LN229 cells cultured with CM from HMC3 cells overexpressing lnc-TALC after subsequent TMZ treatment for 24 hours (Fig. 5D; Supplementary Fig. S9C). Comet assays showed that C5aRa caused more TMZ-induced DNA damage than TMZ alone in both LN229 and HG7 cells (Fig. 5E). IF assays revealed a lower level of expression of RAD51 and higher level of expression of γH2AX in LN229 cells treated with TMZ combined with C5aRa than in LN229 cells treated with TMZ alone (Fig. 5F and G). In addition, the rates of apoptotic cells in the TMZ combined with C5aRa group were increased compared with those in the TMZ-alone group (Fig. 5H and I). We also estimated the survival of patients with GBM using TCGA data and found that C5 significantly correlated with a poor prognosis (P = 0.0396 in TCGA–RNA-seq, P = 0.0349 in TCGA–Agilent, and P = 0.0051 in TCGA–HGU133a; Fig. 5J). Furthermore, the patients with lower C5 expression had a longer median overall survival time than those with higher C5 expression (Supplementary Table S10).
Our previous results demonstrated that exosome-transmitted lnc-TALC enhanced the release of C5a by microglia via activation of the p38 MAPK signaling pathway. CM from lnc-TALC–overexpressing HMC3 cells treated with SB203580 attenuated levels of pATM, pATR, and RAD51 and increased levels of γH2AX and DDR in LN229 cells after subsequent TMZ treatment for 24 hours when compared with CM from lnc-TALC–overexpressing HMC3 cells treated with DMSO (Supplementary Fig. S10A and S10B). FC and IF assays showed that CM from lnc-TALC–overexpressing HMC3 cells treated with SB203580 compared with CM from lnc-TALC–overexpressing HMC3 cells treated with DMSO increased TMZ-induced DNA damage and the apoptotic rate in LN229 cells (Supplementary Fig. S10C–S10F). Overall, these results demonstrated that microglia-derived C5a induced by lnc-TALC increased DDR contributing to TMZ resistance in GBM cells.
C5a receptor antagonism enhances TMZ sensitivity in vivo
To examine the effect of C5 release induced by lnc-TALC in vivo, we established an orthotopic mouse model of GBM via inoculation of GL261 scramble cells, GL261 cells overexpressing lnc-TALC or 261R cells into C57BL/6 mice and treated the mice with TMZ and/or C5aRa (Fig. 6A). Bioluminescence imaging showed that GBM derived from GL261 cells overexpressing lnc-TALC or 261R cells exhibited limited inhibition by TMZ, but TMZ combined with C5aRa treatment significantly inhibited growth of these tumors (Fig. 6B and C). Mice bearing GBM derived from GL261 cells overexpressing lnc-TALC or 261R cells and treated with the combination had prolonged survival compared with those treated with TMZ alone (Fig. 6D).
IHC results showed that C5a expression was significantly decreased in mice bearing GBM derived from the scramble cells compared with those from GL261 cells overexpressing lnc-TALC in control, TMZ, and the combined treatment groups (Fig. 6E; Supplementary Fig. S11A). However, the combined treatment group had increased γH2AX expression in mice bearing GBM derived from GL261 cells overexpressing lnc-TALC (Supplementary Fig. S11B). In addition, we performed IF assays to detect the expression levels of p38 phosphorylation and C5 expression in microglia in vivo. With TMZ treatment alone, tumor-associated microglia in GBM derived from GL261 scramble cells had lower levels of p38 phosphorylation and C5 expression than those in GBM derived from 261R cells or GL261 cells overexpressing lnc-TALC. Furthermore, the combined treatment produced decreased levels of p38 phosphorylation and C5 expression in tumor-associated microglia in GBM derived from GL261 scramble cells compared with those in GBM derived from 261R cells or GL261 cells overexpressing lnc-TALC (Supplementary Fig. S12A and S12B). The safety of C5aRa was confirmed, as no pathologic changes in the visceral organs, hematologic toxicity, and significant impacts on hepatic and renal functions were observed (Supplementary Fig. S13; Supplementary Tables S11 and S12). Overall, these data demonstrate that blockade of lnc-TALC–mediated cross-talk between GBM cells, and microglia enhanced TMZ sensitivity in a mouse GBM model (Fig. 7).
Current standard of care for patients with GBM consists of surgical resection followed by concurrent radiotherapy and chemotherapy, but GBM patients still have a poor prognosis (28). In our previous studies, we explored the specific immunologic signature of GBM, providing evidence for potential targeted immunotherapeutic strategies (29–32). TAMs are important regulators in the TME, affecting matrix components, immune evasion, and angiogenesis, and thereby influencing tumor growth and proliferation (33). In the GBM microenvironment, TAMs predominantly originate from tissue-resident and reactive microglia (34). With advancing tumor grade, the increase in infiltrating microglia in the GBM microenvironment is accompanied by enhanced proliferation of malignant cells and larger pools of undifferentiated GBM cells (35). A deeper understanding of the mechanisms by which microglia affect the GBM microenvironment has the potential to uncover new therapeutic strategies to improve the prognosis of patients with GBM.
Tumor cell–derived exosomes influence noncancer cells to generate a TME that permits for tumor growth and metastasis via their cargos, including oncoproteins and oncopeptides, RNA species, lipids, and DNA fragments (36, 37). In our previous study, we found that lnc-TALC plays a key role in TMZ resistance. However, whether lnc-TALC affects the microglia in the GBM microenvironment was not investigated. In this study, we isolated and purified exosomes from the supernatants of GBM cells overexpressing lnc-TALC and scramble cells. There was a higher level of expression of lnc-TALC in exosomes derived from GBM cells overexpressing lnc-TALC than in exosomes derived from scramble cells. Prior research indicates that exosomes have a substantial effect on macrophages/microglia in the TME (38). Herein, we observed that biological markers of M2 microglia, including cytokines such as TGFβ, IL4, and IL10, remained elevated after culture with exosomes from GBM cells overexpressing lnc-TALC. Our results suggest that exosomes derived from GBM cells contribute to the malignant phenotype of GBM cells by transferring lnc-TALC to microglia, promoting M2 polarization and the generation of an immunosuppressive GBM microenvironment.
LncRNAs are a diverse class of RNAs that engage in numerous biological processes across many diseases (39). To investigate how exosome-transmitted lnc-TALC regulates the polarization of microglia to the M2 phenotype, we used ChIRP-MS and RIP-PCR assays and identified ENO1 as a lnc-TALC-binding protein. ENO1 has many characteristics and biological functions and is involved in many diseases, including the occurrence and metastasis of cancer (40). As a metabolic enzyme involved in the synthesis of pyruvate, ENO1 is a potential target of novel immunotherapies (41). ENO1 is associated with GBM progression, and knockdown of ENO1 expression suppresses cell growth, migration, and invasion progression by inactivating the PI3K/Akt pathway in GBM cells (42). In our study, we performed GO pathway analysis to elucidate the potential role of ENO1 in microglia and found that ENO1 was correlated with the MAPK signaling pathway. ENO1 activates the AMPK pathway to regulate the malignant phenotype of pulmonary artery smooth muscle cells (43). Moreover, senescent human T cells can spontaneously engage the metabolic master regulator AMPK to trigger recruitment of p38 to the scaffold protein TAB1, which causes autophosphorylation of p38 (44). We found that inhibition or knockdown of ENO1 significantly decreased the level of phosphorylated p38 in microglia overexpressing lnc-TALC, indicating that lnc-TALC could promote the phosphorylation of p38 by binding ENO1 in microglia.
To further explore the role of the lnc-TALC/ENO1/p38 axis in the communication between GBM cells and microglia, we performed exocrine protein gene-expression profiling of monocytes and found that the level of expression of the C5 gene was downregulated by p38 inhibition. In the pathologic conditions of autoimmune diseases, the abnormal activation of various signaling pathways, including the p38 MAPK pathway, leads to a global system breakdown induced by disorder of the complement cascade, such as the macrophage activation syndrome and the iatrogenic “cytokine storm” (45). C5 is a component of the complement system, which plays key roles in pathogen immunosurveillance and tissue homeostasis (46). Research indicates that C5 is a critical component of humoral immunity and is implicated in the development of many cancers (47). The transactivation activity of MEF2C, a p38 target, is mediated by p38 through increasing phosphorylation of Thr293, Thr300, and Ser387 (26). Our results further revealed that MEF2C had two identical binding sites in the promoter region of the C5 gene and promoted the transcriptional activity of the C5 gene.
Another study indicated the feasibility of using complement biomarkers for cancer diagnosis and complement inhibitors for cancer treatment (48). uPA+ macrophages regulate C5a generation during premalignant progression, which in turn fosters squamous carcinogenesis (25). We revealed that C5a could be activated by uPA in microglia cultured with exosomes derived from GBM cells overexpressing lnc-TALC. Our results showed that C5a administration elevated the expression levels of DDR-related proteins, increasing TMZ resistance in GBM cells. Complement-targeted C5aRa blocks C5a-induced receptor internalization to produce therapeutic action (49). It has been reported that C5aRa can cross the blood–brain barrier to maintain the inflammatory cytokine balance, improve adult neurogenesis in the hippocampus, and relieve sleep deprivation (50). In the present study, C5aRa attenuated the high levels of DDR-related proteins induced by TMZ treatment in GBM cells cultured with CM from microglia overexpressing lnc-TALC. To explore whether inhibition of the p38-activated C5 release by microglia conferred by exosomal lnc-TALC would make tumors sensitive to chemotherapy, we established an orthotopic mouse model of GBM and administered TMZ and/or C5aRa. The reduced TMZ resistance and prolonged lifespan of the mice treated with TMZ combined with C5aRa revealed that blockade of the lnc-TALC–mediated cross-talk between GBM cells and microglia enhanced TMZ sensitivity in the mouse GBM model. The safety of such a combination therapy was also confirmed in the mice. Neither pathologic changes in the visceral organs nor marked effects on hepatic and renal functions were found, indicating that the combination treatment might serve as a novel, efficacious and safe therapeutic strategy for GBM treatment.
Precision treatment strategies are needed to improve the prognosis of glioma patients (51, 52). It was reported that lncRNAs played an important role in radiotherapy resistance (53). The lncRNAs linc-RA1 and AHIF contribute to glioma radioresistance by regulating oncogenic signaling pathways (54, 55). The role of lnc-TALC in glioma radioresistance remains unclear. However, the activated ENO1 induced by lnc-TALC promoted the activity of the p38 MAPK pathway and had an influence on the malignant phenotype of glioma cells (56). In various cancers, the p38 MAPK signaling pathway is correlated with radioresistance (57–59). In the GBM microenvironment, C5a secreted from stromal cells activates the p38 signaling pathway in tumor cells (60), and is a key player in the tumor response to radiotherapy (61). However, the potential role of the lnc-TALC/ENO1/p38 pathway in radiotherapy resistance in GBM needs to be further explored.
In summary, this study revealed that exosome-transmitted lnc-TALC could regulate microglial M2 polarization and promote TMZ resistance in GBM cells through C5a release induced by the p38 MAPK signaling pathway via binding to ENO1 in microglia. The C5aRa immunotherapy significantly attenuated lnc-TALC–mediated TMZ resistance in GBM. Thus, our present work deepens understanding of the interaction between GBM cells and the TME, and suggests that blockade of lnc-TALC–mediated cross-talk between GBM cells and microglia might be a novel therapeutic strategy.
No disclosures were reported.
Z. Li: Data curation, formal analysis, validation, visualization, writing–review and editing. X. Meng: Data curation, software, visualization, writing–review and editing. P. Wu: Validation, methodology. C. Zha: Validation, methodology. B. Han: Validation, methodology. L. Li: Software, validation, methodology. N. Sun: Validation, methodology. T. Qi: Investigation, visualization. J. Qin: Validation, methodology. Y. Zhang: Data curation, validation. K. Tian: Software, methodology. S. Li: Software, methodology. C. Yang: Data curation, software. L. Ren: Data curation, software. J. Ming: Methodology, writing–review and editing. P. Wang: Software, validation. Y. Song: Software, validation. C. Jiang: Resources, data curation, supervision, project administration. J. Cai: Resources, data curation, formal analysis, supervision, validation, project administration, writing–review and editing.
This work was supported by the National Natural Science Foundation of China (No. 81874204, No. 81772666, No. 81972817, No. 82073298, and No. 82003022), the Heilongjiang Provincial Key R & D Project (GA21C002), the Central Government Supporting Local University Reform and Development Fund for Excellent Youth Talents (0202-300011190006), Karolinska Institutet Research Foundation Grants 2020–2021 (No. FS-2020:0007), the China Postdoctoral Science Foundation (2019M660074 and 2020T130157), the Heilongjiang Postdoctoral Science Foundation (LBH-Z18103 and LBH-Z19029), the Heilongjiang Health and Family Planning Commission Foundation (2018372 and 2019-102), and the Harbin Medical University Scientific Research Innovation Fund (YJSKYCX2019-48HYD).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.