Abstract

   The tumor microenvironment (TME) of gliomas comprises glioma cells and
   surrounding cells, such as astrocytes, macrophages, T cells, and
   neurons. In the TME, glioma cells can activate normal human astrocytes
   (NHAs) through the secretion of exosomes and the activation of
   astrocytes can further improve the progression of glioma, leading to a
   poor prognosis for patients. However, the molecular mechanisms
   underlying NHAs activation by gliomas remain largely unknown. It this
   study, glioma-derived exosomes (GDEs) play an important role in the
   modulation of autophagy and activation of NHAs. Compared with normoxic
   GDEs, hypoxic glioma-derived exosomes (H-GDEs) further improved
   autophagy and activation of astrocytes, which strongly promoted the
   progression of glioma cells. In an miRNA array between two types of
   exosomes from gliomas, miR-423-3p was highly expressed in H-GDEs and
   played an important role in autophagy, resulting in the activation of
   NHAs. The mechanism by which hypoxic glioma cells react with NHAs to
   create an immunosuppressive microenvironment was identified and
   15d-PGJ2 was established as an effective inhibitor of miR-423-3p to
   suppress NHAs activation. These findings provide new insights into the
   diagnosis and treatment of gliomas by targeting autophagy and
   miR-423-3p expression.

   graphic file with name 41419_2025_7576_Figa_HTML.jpg

   Subject terms: Cancer microenvironment, Cancer in the nervous system,
   miRNAs

Introduction

   Glioma is the most malignant and aggressive type of primary brain
   tumor, with a median survival of 12–15 months in adults [[46]1, [47]2].
   Despite recent advances in therapeutic strategies for glioblastoma
   multiforme (GBM), incorporating multimodal techniques, such as surgery,
   chemotherapy, and radiotherapy, the treatment of glioma is still very
   limited owing to its aggressive invasiveness and high recurrence rate
   [[48]3]. Formation of an immunosuppressive microenvironment composed of
   endothelial cells, neurons, and astrocytes is largely influenced by
   glioma cells. Divergent cells within the tumor microenvironment (TME)
   synchronize to increase the proliferation, invasion, immune
   suppression, and angiogenesis of glioma cells [[49]4, [50]5].
   Therefore, interfering with the crosstalk between tumor cells and TME
   has become a novel strategy to address the growth and recurrence of
   gliomas.

   Hypoxia caused by mismatched growth between blood vessels and the
   glioma volume is considered responsible for the malignant behavior of
   gliomas [[51]6]. Hypoxia increases the resistance [[52]7], stemness,
   invasion, and angiogenesis of gliomas [[53]8]. Under hypoxic
   conditions, similar to the influence of irradiation and temozolomide,
   hypoxia can induce normal astrocytes to become reactive astrocytes
   (RAs) [[54]9], indicating that hypoxia is an important factor in
   promoting TME progression.

   Astrocytes are the most abundant type of glial cell in the brain and
   play an important role in the maintenance of neuronal function. The
   ability of astrocytes to proliferate extensively in response to
   injuries and insults has led to the hypothesis that astrocytes could
   serve as the cells of origin for gliomas. Astrocytes can be
   differentiated into tumor glial cells or tumor-origin cells by
   overexpressing oncogenes in vivo and in vitro [[55]4]. In addition, RAs
   can induce the release of O6-alkylguanine DNA alkyltransferase mRNA
   from cells via exosomes. These exosomes can be absorbed into glioma
   cells by endocytosis to block temozolomide (TMZ)-induced apoptosis and
   improve their TMZ resistance [[56]10, [57]11]. Accumulating evidence
   has confirmed that RAs promote glioma progression; however, whether
   gliomas contribute to astrocyte activation remains unclear.

   Exosomes are cell-secreted, vesicle-like corpuscles 30–120 nm in
   diameter. These vesicles deliver various components, such as proteins
   and RNA, to target cells via fluid circulation. Tumor cells can
   communicate with distant cell types by secreting exosomes to transfer
   genetic material and create an immunosuppressive TME [[58]12, [59]13].
   Under hypoxic conditions, tumors increase their content and enhance the
   secretion of exosomes, which subsequently exert detrimental effects on
   recipient cells [[60]14–[61]16]. Notably, RAs can release exosomes into
   glioma cells and increase their chemical resistance [[62]10].
   Therefore, exosomes play an important role in mediating the crosstalk
   between gliomas and astrocytes.

   MicroRNAs (miRNAs) are endogenous small RNAs (20–24 nucleotides)
   [[63]17]. These molecules have many regulatory effects on cells, and
   beneficial effects on glioma proliferation and progression [[64]3,
   [65]18]. In addition, miRNAs induce M2-like polarization in peritumoral
   macrophages, indicating their important role in the TME.

   Autophagy is a process in which lysosomes phagocytose cytoplasmic
   proteins and organelles to form autophagosomes, which then degrade
   their contents [[66]19]. Autophagy can protect cells from death under
   adverse circumstances, such as hypoxia, starvation, and chemotherapy
   [[67]20]. Despite its beneficial role under physiological conditions,
   dysregulation of autophagy promotes the proliferation and growth of
   tumors. In pancreatic ductal adenocarcinoma, tumor cells specifically
   promote autophagy in pancreatic stellate cells (PSCs) and induce the
   degradation of proteins in PSCs to produce alanine, which can be taken
   up by tumor cells to promote their metabolic activity [[68]21].
   Furthermore, in gliomas, tumor cells can induce M2 macrophage
   polarization via enhanced autophagy [[69]22]. Therefore, tumor cells
   can induce autophagic dysregulation in the TME to facilitate their
   growth.

   15-deoxy-(Delta12,14)-prostaglandin J2 (15d-PGJ[2]) is a selective
   peroxisome proliferator-activated receptor gamma (PPARγ) agonist and a
   covalent peroxisome proliferator-activated receptor δ (PPARδ) agonist.
   It inhibits tumor growth through its pro-apoptotic and anti-angiogenic
   effects. In glioma, 15d-PGJ[2] induces A172 glioma cell death by
   increasing reactive oxygen species (ROS) and causing mitochondrial
   dysfunction [[70]23], highlighting its potential in glioma treatment.
   Additionally, in the TME, 15d-PGJ[2] suppresses angiopoietin-II and
   inflammatory cytokines (IL-1β, TNF-α, and TGF-β) by activating PPARγ
   and producing ROS, thereby modulating tumor vasculature [[71]24].
   Therefore, 15d-PGJ[2] may also influence astrocyte activation in the
   glioma microenvironment, though the underlying mechanisms require
   further investigation.

   In this study, whether GDEs contribute to the activation of normal
   astrocytes in the TME was investigated and the underlying mechanisms
   were explored. GDEs induce autophagy and activate normal human
   astrocytes (NHAs). Compared with GDEs under normal conditions (N-GDEs),
   hypoxic GDEs (H-GDEs) induce stronger activation of NHAs. Next,
   upregulated miRNAs in H-GDEs were compared to N-GDEs and found that
   only miR-423-3p in H-GDEs markedly induced autophagy in NHAs and
   subsequently promoted their activation. Moreover, the downstream
   cellular mechanisms were explored and LAMP3 was identified as an
   important molecule that mediates miR-423-3p-induced autophagy in NHAs.
   Using the Connectivity Map (CMap) database,
   15-deoxy-Δ12,14-prostaglandin J2, a selective and covalent PPARδ
   agonist, was identified as a potent inhibitor capable of suppressing
   the activation of NHAs.

Materials and methods

Cell lines and culture

   The glioma cell line, U251, was purchased from the Chinese Academy of
   Sciences Cell Bank. NHAs and the primary human GBM biopsy-propagated
   tumor cell line, P3#GBM, were kindly provided by Department of
   Biomedicine, University of Bergen, Norway. NHAs and U251 were cultured
   in Dulbecco’s modified Eagle’s medium (Thermo Fisher Scientific,
   Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS,
   Vivacell, Shanghai, China). P3#GBM cells were cultured in neurobasal
   medium (NBM, Gibco™, Waltham, MA, USA) with 2% B27 supplement (Thermo
   Fisher Scientific), 1% GlutaMAX (Thermo Fisher Scientific), 1/5000
   recombinant human epidermal growth factor (Novoprotein, Shanghai,
   China), and 1/5000 recombinant human fibroblast growth factor 2
   (Novoprotein).

Exosome isolation

   U251 cells were cultured in DMEM with 10% exosome-free FBS and P3#GBM
   cells were cultured in NBM as described above. Both cell types were
   cultured under normoxic (21% O[2]) and hypoxic (1% O[2]) conditions in
   a CB60 incubator (Binder GmbH, Tuttlingen, Germany). The supernatant
   was collected every 2 days, and exosomes from cells and blood were
   isolated using gradient centrifugation (2000 × g, 10 min; 5000 × g,
   10 min) and ultracentrifugation (100,000 × g, 1 h) [[72]25]. The
   collected exosomes were stored at −80 °C. The concentration of exosomes
   was determined using a BCA Protein Assay kit (Beyotime, Shanghai,
   China) according to the manufacturer’s instructions and the
   concentration of 0.1 μg/ml [[73]19, [74]26, [75]27] was used to treat
   the NHAs.

   Other materials and methods used in this study are provided in the
   [76]Supplementary Materials.

Results

H-GDEs induce autophagy in NHAs in vitro

   To examine the effects of GDEs on the function of astrocytes, GDEs from
   U251 and P3#GBM cells cultured under normoxic (N-GDEs) and hypoxic
   (H-GDEs) conditions were extracted. Transmission electron microscopy
   (TEM) showed that the contents derived from the supernatant were
   rounded objects with diameter of 30–100 nm (Fig. [77]1A), indicating
   the successful extraction of exosomes. Moreover, western blotting
   revealed an enriched expression of exosomal markers, including CD9 and
   TSG101, in the extracted vesicles, whereas calnexin was not detected
   (Fig. [78]1B). When co-cultured with NHA, exosomes labeled with PKH67
   were observed in the cytoplasm of NHA cells, suggesting that exosomes
   could be absorbed by NHA cells (Fig. S[79]1A). The exosome
   concentration was determined using nanoparticle tracking analysis with
   ZetaView (Particle Metrix; Fig. S[80]1B).

Fig. 1. H-GDEs induce autophagy in NHAs in vitro.

   [81]Fig. 1
   [82]Open in a new tab

   A TEM images of exosomes derived from U251 and P3#GBM under normoxia
   and hypoxia, respectively (scale bar, 100 μm). B Western blotting was
   used to detect the expression of CD9, tsg101, and calnexin in protein
   derived from P3#GBM, N-GDEs, and H-GDEs. Exosomes were extracted from
   U251 and P3#GBM under normoxia and hypoxia, respectively. C, D TEM
   Images Illustrating autophagosome formation in NHAs treated with PBS,
   N-GDEs, and H-GDEs from U251 and P3#GBM cells for 48 h. Autophagosomes
   are indicated by arrowheads. Representative images and statistical data
   on autophagosome counts (scale bar, 1.2 μm). E Western blotting was
   used to detect the expression levels of SQSTM1-P62, MAP1LC3B, and GAPDH
   in NHAs treated with PBS, N-GDEs, and H-GDEs derived from U251 and
   P3#GBM cells for 48 h. F–H Immunofluorescence staining for MAP1LC3B
   (green) to measure the level of autophagy in NHAs treated with PBS,
   N-GDEs, and H-GDEs derived from P3#GBM for 48 h; 3-MA and siBecn1 were
   added to the H-GDEs to inhibit autophagy and cell nuclei were stained
   with DAPI (blue). Representative images of three sets of experiments
   (scale bar, 25 μm). I, J TEM images illustrating autophagosome
   formation in NHAs treated with PBS, siNC, and H-GDEs derived from
   P3#GBM cells for 48 h; 3-MA and siBecn1 were added to inhibit
   autophagy. Autophagosomes are indicated by arrowheads. Representative
   images and statistical data on autophagosome counts are shown. K
   Western blotting was used to detect the expression level of SQSTM1-P62,
   MAP1LC3B, and GAPDH in NHAs treated with PBS, siNC, and H-GDEs derived
   from P3#GBM for 48 h; 3-MA and siBecn1 were added to the H-GDEs-treated
   group to inhibit autophagy. (Data reflects the mean ± SEM.*P < 0.05;
   **P < 0.01; ***P < 0.001; n = 3). GBM glioblastoma multiforme, H-GDE
   hypoxic glioma-derived exosome, N-GDE normoxic glioma-derived exosome,
   NHA normal human astrocyte, TEM transmission electron microscopy.

   Because of the hypothesis that defective autophagy triggers reactive
   astrocytes, therefore, the effects of GDEs on autophagy in astrocytes
   were investigated. The N-GDEs and H-GDEs collected from U251 and P3#GBM
   cells, and PBS were added to the culture media of NHAs. TEM was used to
   measure the formation of autophagosomes. The volume of autophagosomes
   significantly increased after H-GDE treatment compared with that after
   N-GDE treatment (Fig. [83]1C, D). The expression of the
   autophagy-related molecules, LC3B and P62, were detected in NHAs using
   western blotting (Fig. [84]1E). Both N-GDEs and H-GDEs increased
   autophagy in NHAs, with a greater increase in astrocytes treated with
   H-GDEs, indicating that GDEs directly affect autophagy in astrocytes.
   Similarly, LC3B within astrocytes increased upon exposure to H-GDEs
   sourced from U251 (Fig. S[85]2A) and P3#GBM (Figs. [86]1F, S[87]2C)
   based on immunofluorescent staining. The inhibition of autophagy using
   3-methyladenine (3-MA) attenuated the expression of LC3B in
   H-GDE-treated NHAs, further confirming that H-GDEs promote autophagy in
   astrocytes (Figs. [88]1G, S[89]2D). Additionally, siRNA-mediated
   knockdown of beclin1 was conducted and the efficacy of this knockdown
   was validated using western blotting (Fig. S[90]2A; si-1 was used in
   the following experiments) as an alternative approach to inhibit
   autophagy. Consistent with the 3-MA treatment, the knockdown of beclin1
   resulted in a similar downregulation of LC3B (Figs. [91]1H, S[92]2D).
   Notably, both 3-MA and beclin1 silencing effectively reversed
   H-GDE-induced autophagosome assembly (Fig. [93]1I, J). Furthermore,
   western blotting revealed the anticipated downregulation of LC3B and
   concurrent upregulation of P62 following treatment with these
   inhibitors (Fig. [94]1K). Collectively, these findings highlight the
   ability of H-GDEs to induce autophagy in astrocytes.

H-GDEs induce a stronger activation of NHAs than that with N-GDEs

   Given that GDEs can induce autophagy in NHA, the subsequent effects of
   GDEs on astrocyte function was determined. Using EdU and Transwell
   assays, we found that both N-GDEs and H-GDEs enhanced the proliferation
   and migration of NHAs, with H-GDEs exhibiting stronger effects than
   N-GDEs (Fig. [95]2A, C, E). Enzyme-linked immunosorbent assay (ELISA)
   showed that the increase in the proliferation and migration of NHAs was
   accompanied by a significant increase in the secretion of IL-6 and IL-8
   in H-GDE-treated cells (Fig. [96]2I), indicating the transformation of
   NHAs into reactive astrocytes [[97]28, [98]29] by H-GDEs. GFAP is a
   canonical marker of reactive astrocytes [[99]30], therefore, GFAP
   expression was assessed after treatment with N-GDEs and H-GDEs. The
   findings revealed pronounced upregulation of GFAP expression in NHAs
   following exposure to H-GDEs (Fig. [100]1G), substantiating the potent
   pro-reactive role of H-GDEs in astrocyte activation.

Fig. 2. H-GDEs promoted a higher level of NHA activation than N-GDEs.

   [101]Fig. 2
   [102]Open in a new tab

   A, B Migration of cells passing through the Transwell chambers of NHAs
   treated with PBS, N-GDEs, and H-GDEs derived from U251 and P3#GBM cells
   was measured using microscopy. siBecn1 and 3-MA were used to inhibit
   the autophagy, and representative images are shown (scale bar, 200 μm,
   n = 3). C, D An EdU assay was used to detect the proliferation of NHAs
   treated with PBS, N-GDEs, and H-GDEs derived from U251 and P3#GBM
   cells. siBecn1 and 3-MA were used to inhibit the autophagy, and
   representative images are shown (scale bar, 70 μm, n = 5). E
   Quantitative analysis of migration cell numbers. Presented here is the
   numerical quantification of migrated cell counts captured in (A, B),
   following a 48-h incubation period. F Measurement of EdU assay ratio.
   Presented here is the quantitative evaluation of the ratio of
   EdU-positive cells, as observed in (C, D), following a 48-h incubation
   period. G, H Protein collected from NHAs treated with PBS, siNC,
   N-GDEs, and H-GDEs derived from U251 and P3#GBM cells was used to
   detect the expression of GFAP and GAPDH through western blotting to
   detect the activation of astrocytes; siBecn1 and 3-MA were used to
   inhibit the autophagy. Numeric annotations below each blot represent
   the mean GFAP expression values from three replicate experiments,
   quantified by grayscale analysis. Levels are normalized to a control
   set at an arbitrary value of 1. I, J IL-6 and IL-8 ELISA kits were used
   to measure the secretion ability of NHAs treated with PBS, siNC,
   N-GDEs, and H-GDEs derived from U251 and P3#GBM cells; siBecn1 and 3-MA
   were used to inhibit autophagy. The statistical results are shown
   (n = 3). (Data reflects the mean ± SEM. *P < 0.05; **P < 0.01;
   ***P < 0.001). GBM glioblastoma multiforme, H-GDE hypoxic
   glioma-derived exosome, N-GDE normoxic glioma-derived exosome, NHA
   normal human astrocyte.

   GDEs promoted both autophagy and the transformation of NHAs to RAs.
   Next, whether autophagy played a role in mediating this transformation
   was investigated. Inhibition of autophagy by 3-MA or knocking down of
   beclin1 in the H-GDE-treated group attenuated the H-GDE-induced
   secretion of IL-6 and IL-8 (Fig. [103]2J). Whether the functional
   changes observed in NHAs upon GDEs exposure were mediated by autophagy
   was then investigated. EdU and Transwell assays also showed that the
   downregulation of autophagy blocked the effects of H-GDEs on the
   proliferation and migration of NHAs (Fig. [104]2B, D, F). Additionally,
   western blotting revealed that autophagy inhibition, achieved either
   through 3-MA or genetic knockdown of beclin1, led to a significant
   decrease in GFAP expression (Fig. [105]2G). Collectively, these results
   strongly suggest that H-GDEs can induce the transformation of NHAs into
   a reactive phenotype by regulating autophagy.

H-GDEs induce autophagy in NHAs by miR-423-3p and the effect can be inhibited
by 3-MA

   Exosomes can transport miRNAs from cell to cell [[106]18]. To determine
   the miRNA responsible for the GDE-induced transformation of NHAs, the
   miRNA array was used to analyze the miRNA contents in N-GDEs and H-GDEs
   derived from U251 and P3#GBM cells (Fig. [107]3A). The differentially
   expressed miRNAs in N-GDEs were compared with those in H-GDEs and the
   top 10 miRNAs that were highly expressed in H-GDEs extracted from both
   cell lines were identified (Fig. [108]3B, P < 0.001, fold-change > 1).
   The effects of the 10 miRNAs on the regulation of autophagy in NHAs
   were determined. MiR-423-3p significantly elevated the expression of
   LC3B in NHAs, as measured semi-quantitatively (Fig. [109]3C, D),
   indicating its important role in modulating autophagy. Quantitative
   real-time PCR (qRT-PCR) was performed on N-GDEs and H-GDEs derived from
   U251 (Fig. S[110]3A) and P3# GBM cells (Fig. [111]3E), which confirmed
   the upregulation of miR-423-3p. Additionally, the expression level was
   higher in the exosomes collected from the blood of patients with GBM
   compared with those collected from healthy patients (Fig. [112]3F). To
   further test whether miR-423-3p regulates autophagy in NHAs, 3-MA was
   used to inhibit autophagy in NHAs transfected with miR-423-3p mimics.
   qRT-PCR was performed on miR-423-3p-transfected NHAs to validate the
   increase in miR-423-3p expression (Fig. S[113]3B). In addition, miRDB
   ([114]http://www.mirdb.org/) and TargetScan
   ([115]https://www.targetscan.org/) were used to predict the potential
   downstream target genes of miR-423-3p. Together with a previous study
   [[116]31], both predictions showed that cytoplasmic poly(A)-binding
   protein-1 (PABPC1) and RAP2C, members of the RAS oncogene family, are
   downstream target genes of miR-423-3p. Therefore, qRT-PCR was conducted
   to quantify the expression of both genes in the miR-423-3p-transfected
   group (Fig. S[117]3C), which further confirmed successful transfection.
   Western blotting and immunofluorescence images showed that miR-423-3p
   increased autophagy in NHAs (Figs. [118]3G, I, and S[119]4A). The same
   trend was observed when beclin1 was knocked down in
   miR-423-3p-transfected cells (Figs. [120]3H, I and S[121]4B),
   indicating the autophagic induction effect of miR-423-3p.

Fig. 3. H-GDEs induced autophagy in NHAs by miR-423-3p and could be inhibited
by 3-MA.

   [122]Fig. 3
   [123]Open in a new tab

   A MiRNA array for normoxic and hypoxic exosomes derived from U251 and
   P3#GBM cells; the upregulated miRNAs are shown as heat map. B Venn
   diagram for the intersection of differences in exosomes in both cell
   lines above. C, D Western blot for candidate miRNAs inducing autophagy
   intersected above. Representative images of three sets of experiments
   and the statistical results of relative grayscale ratio of LC3B
   II/GAPDH are shown. E A qRT-PCR analysis to detect the relative
   expression of miR-423-3p in GDEs from P3#GBM. F A qRT-PCR analysis to
   detect the relative expression of miR-423-3p in exosomes derived from
   the patients with GBM and healthy patients. Statistical results of
   three groups are shown. G, H Immunofluorescence staining for MAP1LC3B
   (green) to measure the level of autophagy in NHAs transfected with
   miR-control and miR-423-3p; 3-MA and siBecn1 were used to inhibit
   autophagy in NHAs and cell nuclei were stained with DAPI (blue).
   Representative images of three sets of experiments are shown (scale
   bar, 25 μm). I NHAs were transfected with miR-control and miR-423-3p;
   3-MA and siBecn1 were used to inhibit autophagy in NHAs. The relative
   expression of SQSTM1-P62, MAP1LC3B, and GAPDH in NHAs was detected
   using western blotting. J, K Immunofluorescence staining for MAP1LC3B
   (green) to detect the level of autophagy in NHAs treated with PBS,
   inhibitor control, N-GDEs, and H-GDEs derived from P3#GBM; miR-423-3p
   plasmid was transfected into P3#GBM to get miR-423-3p overexpression
   exosomes under normoxic conditions. miR-423-3p inhibition RNA was
   transfected to inhibit miR-423-3p in NHAs and cell nuclei were stained
   by DAPI (blue). Representative images of three sets of experiments are
   shown (scale bar, 25 μm). L NHAs were treated with PBS, inhibitor
   control, N-GDEs, and H-GDEs derived from P3#GBM; miR-423-3p plasmid was
   transfected into P3#GBM to get miR-423-3p overexpression exosomes under
   normoxic conditions. miR-423-3p inhibition RNA was transfected to
   inhibit miR-423-3p in NHAs. The relative expression of SQSTM1-P62,
   MAP1LC3B, and GAPDH in NHAs was detected using western blotting. M, N
   TEM images illustrating autophagosome formation in NHAs treated with
   miR-control, miR-423-3p, inhibitor control, PBS, N-GDEs, and H-GDEs
   derived from P3#GBM; miR-423-3p plasmid was transfected into P3#GBM to
   get miR-423-3p overexpression exosomes under normoxic conditions.
   miR-423-3p inhibition RNA was transfected to inhibit miR-423-3p in
   NHAs. Autophagosomes are indicated by arrowheads. Representative images
   and statistical data on autophagosome counts are shown. (Data reflects
   the mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001; n = 3). GBM
   glioblastoma multiforme, H-GDE hypoxic glioma-derived exosome, N-GDE
   normoxic glioma-derived exosome, NHA normal human astrocyte, TEM
   transmission electron microscopy.

   Given that miRNA-423-3p can induce autophagy, whether miR-423-3p
   mediates the effects of GDEs on autophagy was investigated. Using
   lentivirus-mediated expression of miR-423-3p in the P3#GBM cells,
   exosomes were extracted from miR-423-3p-overexpressing and negative
   control cells under normoxic conditions. The RNA of the derived
   exosomes was used to perform qRT-PCR assays, which confirmed an
   increase in miR-423-3p expression in the overexpression group (Fig.
   S[124]3D). The extracted GDEs significantly upregulated LC3B in the
   NHAs when compared with that of the controls (Figs. [125]3J, L and
   S[126]4C), and inhibitory RNA against miR-423-3p (Fig. S[127]3E)
   abolished the upregulation of LC3B in the NHAs, as shown using western
   blotting (Fig. [128]3L). Immunofluorescence staining of LC3B showed
   similar results (Figs. [129]3K and S[130]4D). Morphologically,
   miR-423-3p induced the formation of autophagosomes, which was inhibited
   by 3-MA, knockdown of beclin1, and the miRNA inhibitor (Figs. [131]3M,
   N and [132]S4E, F). Collectively, miR-423-3p plays an important role in
   inducing autophagy in NHAs.

MiR-423-3p-induced autophagy in H-GDEs promotes the activation of astrocytes

   MiR-423-3p induce autophagy in NHAs, therefore, whether it promoted
   astrocyte transformation was explored. NHAs were transfected with
   miR-423-3p mimics and treated with or without autophagy inhibition
   using 3-MA or beclin1 knockdown RNA. Transwell and EdU assays showed
   that miR-423-3p significantly increased the proliferation and migration
   of NHAs, which was attenuated by autophagy inhibition (Fig. [133]4A, C,
   E, F). Using ELISA and western blotting, we confirmed that miR-423-3p
   increased the secretion of IL-6 and IL-8 from astrocytes (Fig.
   [134]4I), and upregulated GFAP expression (Fig. [135]4G). These results
   indicate that miRNA-423-3p promotes the transition of NHA to RAs. To
   further examine whether miR-423-3p mediates the effects of GDEs on
   transformation of NHAs, GDEs were extracted from the P3#GBM cells
   overexpressing miR-423-3p under normoxic conditions to treat the NHAs.
   H-GDEs with an miRNA that inhibits miR-423-3p were added to the other
   group. MiR-423-3p significantly enhanced the transformation, whereas
   inhibitory RNA rescued the H-GDE-induced transformation of NHAs (Fig.
   [136]4B, D–F, H, J). Together, these results demonstrate that
   miR-423-3p in H-GDEs promotes the tumor-promoting transformation of
   NHAs through autophagy.

Fig. 4. MiR-423-3p-induced autophagy in H-GDEs promotes the activation of
astrocytes.

   [137]Fig. 4
   [138]Open in a new tab

   A, B Migration of cells passing through the Transwell chambers of NHAs
   treated with miR-control, miR-423-3p, inhibitor control, PBS, N-GDEs,
   and H-GDEs; 3-MA and siBeclin1 were used to inhibit autophagy. The
   miR-423-3p plasmid was transfected into P3#GBM to obtain
   miR-423-3p-overexpressing exosomes under normoxic conditions.
   MiR-423-3p inhibitory RNA was transfected to inhibit miR-423-3p in
   NHAs, and representative images are shown (scale bar, 200 μm, n = 3).
   C, D EdU assay was used to detect the proliferation ability of NHAs
   clarified in (A, B) and representative images are shown (scale bar,
   70 μm, n = 3). E Quantitative analysis of migration cell numbers. The
   numerical quantification of migrated cell counts captured in Figures A
   and B, following a 48-h incubation period. F Measurement of EdU assay
   ratio. Quantitative evaluation of the ratio of EdU-positive cells, as
   observed in (C, D), following a 48-h incubation period. G, H Protein
   collected from NHAs clarified in (A, B) was used to detect the
   expression of GFAP and GAPDH through western blotting to detect the
   activation of astrocytes. Numerical annotations below each blot denote
   the average GFAP expression values from three replicate experiments,
   quantified by grayscale analysis. Expression levels are normalized to a
   control set at 1. I, J Cell supernatant from NHAs clarified in (A, B)
   was collected to detect the secretion of cytokines through IL-6 and
   IL-8 ELISA kits. The statistics were measured by a microplate reader
   (n = 3). (Data reflects the mean ± SEM. *P < 0.05; **P < 0.01;
   ***P < 0.001). GBM glioblastoma multiforme, H-GDE hypoxic
   glioma-derived exosome, N-GDE normoxic glioma-derived exosome, NHA
   normal human astrocyte.

RAs induced by miR-423-3p in H-GDEs promote the progression of glioma in vivo

   Given that glioma cells can induce the transformation of NHAs to RAs
   through exosomal-based mechanisms, whether the transformation of
   astrocytes by GDEs or miR-423-3p can promote the progression of glioma
   was investigated. P3#GBM cells and astrocytes pretreated with PBS,
   N-GDEs, and H-GDEs were co-implanted into the brains of nude mice and
   treated the mice with PBS, N-GDEs, and H-GDEs through the tail vein
   every 3 days. The tumor volume measured by bioluminescence imaging at
   one and 4 weeks post implantation showed that the size of the tumors in
   mice treated with H-GDE was significantly larger than that in mice in
   the other two groups (Figs. [139]5A, B and S[140]5A). Subsequently,
   mice treated with H-GDEs exhibited a shorter survival time than mice in
   the other groups (Fig. [141]5C). H&E and IHC staining of Ki67 showed
   that the tumors in the H-GDE-treated group had few clear borders and
   high Ki67 expression. The astrocyte marker, GFAP, was also stained;
   NHAs treated with H-GDEs exhibited a high expression of heteromorphic
   GFAP, with a clear increase in the number of branches (Figs. [142]5D,
   S[143]5B). Moreover, co-immunostaining of LC3B and GFAP showed that
   H-GDE treatment significantly increased the expression of LC3B and
   abnormal GFAP in NHAs (Fig. [144]5E), suggesting that H-GDEs could
   promote autophagy and transformation in NHAs in vivo.

Fig. 5. RAs induced by miR-423-3p in H-GDEs promoted the progression of
glioma in vivo.

   [145]Fig. 5
   [146]Open in a new tab

   A, B Bioluminescence imaging of tumor-bearing mice treated with P3#GBM
   together with PBS-NHA, N-GDEs-NHA, and H-GDEs-NHA. Representative
   images and statistical analysis of day 28 are shown. C Survival
   analysis for orthotopic xenografts bearing P3#GBM together with
   PBS-NHA, N-GDEs-NHA, and H-GDEs-NHA (five mice per group). D H&E
   staining (scale bar, 300 μm) and IHC staining (scale bar, 100 μm) for
   Ki67 and GFAP from xenograft mice treated with P3#GBM, PBS-NHA,
   N-GDEs-NHA, and H-GDEs-NHA on the day of euthanasia. Representative
   images of three sets of experiments are shown. E Co-immunofluorescence
   staining for LC3B (green) and GFAP (red) in NHAs in orthotopic
   xenografts treated with PBS, N-GDEs, and H-GDEs derived from P3#GBM
   after euthanasia. Cell nuclei were stained with DAPI (blue).
   Representative images of three sets of experiments are shown (scale
   bar, 10 μm). F, G Bioluminescence imaging of tumor-bearing mice treated
   with P3#GBM together with lenti-control-NHA and
   lenti-ov-miR-423-3p-NHA. Representative images and statistical analysis
   of day 28. H Survival analysis for orthotopic xenografts bearing P3#GBM
   together with lenti-control-NHA and lenti-ov-miR-423-3p-NHA (five mice
   per group). I, J H&E staining (scale bar, 300 μm) and IHC staining
   (scale bar, 100 μm) for Ki67 and GFAP from xenograft mice co-planted
   with lenti-control-NHA and lenti-ov-miR-423-3p-NHA. Representative
   images of three sets of experiments and statistical analysis for IHC
   score of Ki67 are shown. K Co-immunofluorescence staining for LC3B
   (green) and GFAP (red) in lenti-control-NHAs and
   lenti-ov-miR-423-3p-NHAs. Cell nuclei were stained with DAPI (blue).
   Representative images of three sets of experiments are shown (scale
   bar, 10 μm). (Data reflects the mean ± SEM. *P < 0.05; **P < 0.01;
   ***P < 0.001; n = 3). GBM glioblastoma multiforme, H-GDE hypoxic
   glioma-derived exosome, N-GDE normoxic glioma-derived exosome, NHA
   normal human astrocyte, RA reactive astrocytes.

   After confirming the role of H-GDEs in facilitating glioma progression
   in vivo, the effects of miR-423-3p on tumor progression was examined.
   NHAs were transferred with lenti-miR-control or lenti-ov-miR-423-3p
   virus, and co-implanted into nude mice alongside P3#GBM cells to
   generate orthotopic xenografts. Bioluminescence imaging revealed that
   by the 4th week, miR-423-3p increased the tumor size in nude mice
   (Figs. [147]5F, G and S[148]5C). Moreover, the survival rate of the
   implanted mice was lower than that of the control mice (Fig. [149]5H).
   Consistently, H&E staining showed strong tumor infiltration in the
   miR-overexpressing group (Fig. [150]5I). Additionally, IHC staining for
   Ki67 and GFAP showed that the activation of NHAs induced by miR-423-3p
   promoted the proliferation rate of gliomas and the expression of
   heteromorphic GFAP in NHAs (Fig. [151]5I, J). Co-immunostaining further
   indicated that lenti-miR-423-3p-expressing mice exhibited higher levels
   of LC3B and heteromorphic GFAP compared with those of control mice
   (Fig. [152]5K). Collectively, these results suggest that miR-423-3p
   derived from H-GDEs can promote the transformation of NHAs in the TME
   into RAs and subsequently facilitate the progression of glioma in vivo.

LAMP3 is the downstream target gene of miR-423-3p and participates in
autophagy in NHAs

   To detect the downstream genes of miR-423-3p involved in its effects on
   autophagy, qPCR array was used to probe the autophagy-related target
   genes in NHAs transfected with the control and miR-423-3p mimics (Fig.
   [153]6A, B). According to previous studies [[154]32–[155]35], lamp3 may
   be a downstream target of miR-423-3p, and its function is closely
   involved in autophagy. The expression of LAMP3 in NHAs treated with
   PBS, N-GDEs, and H-GDEs derived from U251 and P3#GBM cells were
   compared using western blotting. H-GDEs significantly increased the
   expression of LAMP3 in NHAs (Fig. [156]6C). Moreover, in
   miR-423-3p-transfected NHAs, LAMP3 expression increased, and this trend
   was reversed when autophagy was inhibited (Fig. [157]6D). These
   experiments indicated that LAMP3 may be involved in the effects of
   miR-423-3p on the induction of autophagy. A pull-down assay showing a
   direct interaction between miR-423-3p and LAMP3 further supported this
   speculation (Fig. [158]6E).

Fig. 6. LAMP3 is a downstream target gene of miR-423-3p that participates in
the autophagy of NHAs.

   [159]Fig. 6
   [160]Open in a new tab

   A, B NHAs were transfected with negative control mimics and miR-423-3p
   mimics. Utilizing a qPCR array, the heat map and volcano map showcased
   differential gene expression patterns. C, D The relative expression of
   Lamp3 was detected using western blotting in NHAs treated with PBS,
   U251-derived N-GDEs, H-GDEs, P3#GBM-derived N-GDEs, H-GDEs, control
   mimics, siNC, and miR-423-3p mimics; 3-MA and siBeclin1 were added to
   inhibit autophagy respectively. Below each blot, numbers represent the
   mean LAMP3 expression values from three independent experiments,
   quantified using grayscale analysis. Expression levels are normalized
   to a control set at 1. E A pulldown kit was used for NHAs to validate
   the connection between miR-423-3p and Lamp3; expression of human
   antigen R (HUR) was measured in negative and positive proteins (derived
   from the pulldown kit), and the NHA protein was added with
   biotin-marked sense and antisense RNA to confirm the successful
   pulldown of proteins. The expression of LAMP3 and GAPDH was detected in
   NHA protein treated with biotin-marked sense and antisense RNA. F A
   schematic diagram was created using biorender.com. GBM glioblastoma
   multiforme, H-GDE hypoxic glioma-derived exosome, N-GDE normoxic
   glioma-derived exosome, NHA normal human astrocyte.

The 15d-PGJ2 can inhibit the activity of miR-423-3p and suppress the
activation of NHAs

   To identify potential drugs that can inhibit miR-423-3p activity,
   transcriptome sequencing of NHAs treated with miR-control or miR-423-3p
   was conducted, which allowed us to identify differentially expressed
   genes (Fig. [161]7A, B). KEGG enrichment analysis revealed that the
   differentially expressed genes were associated with lysosomal function,
   mitophagy, and metabolic pathways, all of which are associated with
   autophagy (Fig. [162]7C). Additionally, GO enrichment analysis
   highlighted exosomes and focal adhesions, contributing to the
   subsequent formation of the glioma microenvironment (Figs. [163]7D,
   S[164]6A). The CMap database was used to predict potential drugs based
   on the differentially expressed genes. The top three drugs predicted to
   pass through the blood-brain barrier (BBB) were GDC-0879, 15d-PGJ2, and
   linsitinib (Fig. [165]7E). At a concentration of 5 μM, 15d-PGJ2
   exhibited potent inhibitory effects on both glioma cells and exosomes,
   distinguishing its efficacy compared with that of GDC-0879 and
   linsitinib (Fig. [166]7F). Together with the IC50 tested using CCK-8
   (Figs. [167]7G, S[168]7A), the molecular topological polar surface
   area, and molecular weight of these candidates, 15d-PGJ2 was chosen to
   conduct the following experiments.

Fig. 7. 15d-PGJ2 can inhibit the activity of miR-423-3p and suppress the
activation of NHAs.

   [169]Fig. 7
   [170]Open in a new tab

   A Heatmap depicting the cluster analysis of differentially expressed
   genes in NHAs treated with miR-control versus miR-423-3p mimics,
   illustrating distinct gene expression patterns. B Volcano plot
   showcasing differential gene expression in NHAs following treatment
   with miR-control and miR-423-3p mimics (|log2FC| ≥ 1 and P < 0.05). C
   Dot plot representing KEGG pathway enrichment analysis of transcriptome
   sequencing data from NHAs treated with miR-control and miR-423-3p
   mimics, indicating enriched pathways. D Dot plot summarizing the
   cellular component aspect of GO enrichment analysis derived from
   transcriptome sequencing in NHAs treated with miR-control and
   miR-423-3p mimics. E Schematic diagrams of the three top-ranked
   inhibitors of miR-423-3p, commercially available and predicted to cross
   the blood-brain barrier. F qRT-PCR analysis revealing the relative
   expression levels of miR-423-3p in P3#GBM cells and exosomes following
   treatment with 5 μM GDC-0879, 15d-PGJ2 and linsitinib, respectively. G
   Survival curves for NHAs exposed to increasing concentrations 15d-PGJ2
   (0.6–160 μM) over 48 h to determine the IC50 values. H Representative
   images and statistical analysis showing the migration capacity of NHAs
   across Transwell chambers treated with miR-control, miR-423-3p, and
   15d-PGJ2 (scale bar, 200 μm). I Western blotting of protein extracts
   from NHAs treated with miR-Control, miR-423-3p, and 15d-PGJ2,
   demonstrating the expression of GFAP and GAPDH. Relative GFAP levels of
   three replications are quantified via grayscale analysis with control
   set to 1. J EdU assay demonstrates NHAs proliferative response to
   miR-Control, miR-423-3p, and 15d-PGJ2 treatments. Representative images
   and statistical analysis are shown (scale bar, 70 μm) K EdU assay was
   used to detect the proliferation ability of P3#GBM co-cultured with
   NHAs pretreated with miR-control, miR-423-3p, and 15d-PGJ2.
   Representative images and statistical analysis are shown (scale bar,
   70 μm) (Data reflects the mean ± SEM. ns P ≥ 0.05*P < 0.05; **P < 0.01;
   ***P < 0.001; n = 3). GBM glioblastoma multiforme, NHA normal human
   astrocyte.

   15d-PGJ2 is a kind of selective peroxisome proliferator-activated
   receptor gamma (PPARγ) and a covalent PPARδ agonist. It could inhibit
   the proliferation and expansion of glioma [[171]36]. In NHAs, 15d-PGJ2
   treatment led to a decrease in migration rate (Fig. [172]7H), level of
   GFAP (Fig. [173]7I), and proliferation rate (Fig. [174]7J) induced by
   miR-423-3p. Additionally, the activation induced by H-GDEs was also
   inhibited by 15d-PGJ2 in the NHAs (Fig. S[175]7B–D). Subsequently, the
   proliferation rate of P3#GBM cells was observed to decrease when
   co-cultured with NHAs pretreated with 15d-PGJ2 (Fig. [176]7K),
   highlighting the potential anti-proliferative effects of this compound
   on GBM cell growth within a complex cellular microenvironment. Thus,
   15d-PGJ2 was predicted to be a promising inhibitor for decreasing the
   activation of NHAs and formation of the TME.

Discussion

   Although various treatment methods have been used clinically, the
   prognosis of patients with GBM is poor owing to high treatment
   resistance and recurrence rates [[177]37, [178]38]. An important factor
   in drug resistance and recurrence is the TME, which is formed by
   various cells influenced by tumors, such as tumor-associated
   macrophages [[179]39], neurons, astrocytes, and microglia [[180]40].
   Astrocytes, one of the major sources of gliomas according to the
   International Classification of Diseases–Oncology, version 3 and the
   World Health Organization grade [[181]4, [182]41, [183]42], play an
   essential role in remodeling the TME. Indeed, studies has reported that
   astrocytes directly or indirectly regulate the progression and drug
   resistance in gliomas [[184]43, [185]44]. However, whether and how the
   crosstalk between glioma cells and astrocytes contributes to malignant
   changes in the TME remains poorly understood. Exosomes act as mediators
   of intercellular communication in the TME and promote tumor cell
   progression in various cancers [[186]45, [187]46]. Furthermore,
   hypoxia, which results from the disparate growth rates of tumor cells
   and blood vessels, is a key characteristic of gliomas. It promotes
   tumor progression and, more importantly, influences the formation and
   heterogeneity of tumor-derived exosomes, subsequently contributing to
   the development of an immunosuppressive TME [[188]47]. Based on these
   observations, we compared N-GDE and H-GDE effects on the induction of
   RAs. H-GDEs induced a much stronger reactive phenotype in astrocytes,
   accompanied by increased proliferation, migration, cytokine secretion,
   and GFAP expression. Furthermore, when H-GDE-treated astrocytes were
   co-implanted with glioma cells in vivo, a significantly higher
   proliferation rate of glioma than that of the N-GDE group. Consistent
   with previous studies, hypoxia-derived exosomes function as mediators
   of intracellular communication in the TME. For example, Qiu et al.
   reported that H-GDEs led to the increased differentiation and
   activation of myeloid-derived suppressor cells by increasing the
   content of miR-1246 in exosomes [[189]48]. In gliomas, H-GDEs can
   induce the M2-polarization of macrophages by directing IL-6 into
   macrophages [[190]18]. Exosomes have also been reported to induce the
   activation of astrocytes. Exosomes containing miR-21 secreted by
   neurons contributed to the overactivation of astrocytes in a mouse
   model of Alzheimer’s disease [[191]49]. Therefore, the exosomes mediate
   the intracellular communication between glioma and astrocytes and
   transform the detrimental effects of hypoxic conditions on TME
   remodeling. Exosomes participate in message exchange between tumor
   cells and other cell types in the TME by delivering tumor-derived
   contents to recipient cells [[192]50]. Among the genetic contents
   delivered by exosomes, miRNAs have specific and important effects on
   various biological activities of cells, such as proliferation, DNA
   repair, and feedback regulation [[193]51, [194]52]. Therefore, we used
   miRNA arrays to identify potential miRNAs that mediate the effects of
   GDEs on RA formation. After comparing the differentially regulated
   miRNAs between H-GDEs and N-GDEs, miR-423-3p was significantly
   upregulated in H-GDEs. Next, the effect of miR-423-3p on the induction
   of RAs was examined. Similar to the findings of H-GDEs,
   miR-423-3p-transfected NHAs exhibited a significant increase in the
   proliferation, migration, and secretion of proinflammatory cytokines,
   accompanied by elevated levels and atypia of GFAP. Compared to the
   effect of N-GDEs from control cells, N-GDEs with stabilized expression
   of miR-423-3p resulted in a much more reactive phenotype of RAs,
   imitating the effect of H-GDEs. In contrast, when miR-423-3p was
   inhibited in NHAs, the effect of H-GDEs was reduced. Therefore,
   miR-423-3p is a key component that mediates the effects of GDEs on RAs
   induction. The in vitro xenograft model further confirmed these
   findings. The volume of glioma cells co-implanted with astrocytes
   transfected with miR-423-3p was significantly higher than that of mice
   in the control group, leading to a greater reduction in the survival
   time of mice. Consistently, miR-423-3p has been reported to promote the
   progression of gliomas [[195]18] as well as various other tumors,
   including gastric cancer, hepatocellular carcinoma, and colorectal
   cancer [[196]53–[197]55], indicating its functional role in promoting
   malignancy. These results strongly suggest the key role miR-423-3p
   plays in astrocyte activation.

   In addition, the role of miRNAs in neurodegenerative disorders has
   gained considerable attention [[198]56, [199]57]. For instance,
   octadecaneuropeptide alleviates the formation of reactive astrocytes
   induced by 6-hydroxydopamine by downregulating the overexpression of
   miR-21 in models of Parkinson’s disease models [[200]58]. Furthermore,
   downregulation of the miR-106b and miR-101 in patients with Alzheimer’s
   disease results in the expression of amyloid precursor protein and the
   generation of Aβ, resulting in the activation of surface receptors on
   astrocytes and aggravation of Alzheimer’s disease [[201]59]. As for
   miR-423-3p, Zameer et al. elucidated that miR-423-3p is markedly
   upregulated and associated with several high-risk genes linked to
   multiple Parkinson’s disease genes [[202]60]. In amyotrophic lateral
   sclerosis, alterations in the levels of miR-423-3p serve as critical
   biomarkers for the early stages of the disease [[203]61]. These results
   strongly suggest that miR-423-3p plays an important role in the
   occurrence and development of central nervous system diseases.

   Autophagy is an important regulatory process that maintains cellular
   homeostasis. However, in tumor cells, their functions may vary under
   different conditions. In the early stages of tumor formation, autophagy
   can help eliminate tumor cells and inhibit their progression; however,
   when autophagy is dysregulated, it promotes the survival and growth of
   tumor cells [[204]62, [205]63]. Autophagy is involved in the crosstalk
   between tumors and other cell types in the TME [[206]64], suggesting
   that autophagy can also contribute to the formation of a malignant TME.
   The autophagy levels in NHAs treated with H-GDEs and N-GDEs were
   detected. Morphologically, autophagosome formation in astrocytes of the
   H-GDE-treated group increased. The expression level of LC3B in the
   H-GDE group was higher than that in the N-GDE group and the expression
   of P62 was the opposite to that of LC3B, indicating an increase in
   autophagy. Next, whether miR-423-3p is a direct messenger that induces
   autophagy in NHAs was examined. MiR-423-3p induced autophagy in NHAs,
   including the formation of autophagosomes, an increase in LC3B, and a
   decrease in P62. Similar to this study, Zhang et al. reported that the
   upregulation of miR-2188-5p increases the autophagic flux of tracheal
   cells in an ammonia environment [[207]65]. In another study, miR-122
   was shown to induce protective autophagy in hepatocytes under arsenic
   stress [[208]66]. These studies have confirmed the role of miRNAs in
   the regulation of autophagy. However, whether autophagy promotes
   astrocyte activation remained unclear. Using the autophagy inhibitor
   3-MA or knocking of beclin1, NHAs treated with H-GDEs or miR-423-3p
   mimicked the phenotype observed with autophagy inhibition. The typical
   properties of RAs were also attenuated, suggesting that H-GDEs or
   miR-423-3p induced RAs through pathological changes by inducing
   autophagy. However, most published studies suggest that the activation
   of astrocytes is induced by the inhibition of autophagy, which is
   inconsistent with to the results of this study. However, as clarified
   above, autophagy may have different effects on tumors under different
   periods and circumstances. Furthermore, astrocytes are considered one
   of the origins of gliomas. A reactive procedure is required when
   astrocytes are malignantly induced in glioma cells. Therefore, in some
   situations, autophagy in NHAs may be increased to induce RAs, leading
   to tumor progression. In addition, a previous study reported that in
   the TME and distal tissue, tumor cells can induce non-cell-autonomous
   autophagy of normal cells to promote tumor progression in the early
   stages [[209]67], which is consistent with this study. Moreover, most
   studies focusing on the formation of RAs by autophagic inhibition are
   non-oncological. Hence, the effects of autophagy on the TME and
   astrocytes may differ.

   MiRNAs are generally believed to regulate gene expression by inhibiting
   the expression of target genes; however, sometimes, miRNAs can also
   promote the expression of downstream genes. This phenomenon is called
   “miRNA mediated transcriptional activation” or “miRNA enhancement
   effect.” In some cases, certain mRNAs act as a “sponge” to absorb a
   specific miRNA, thereby indirectly increasing the expression of other
   mRNAs that share the same miRNA response elements [[210]68]. Li et al.
   demonstrated that miRNAs alter the chromatin state by interacting with
   chromatin-modifying factors, thereby promoting gene expression
   [[211]69]. Furthermore, miRNAs can directly bind to transcription start
   sites to promote their expression [[212]70]. According to this study,
   miRNAs may interact with proteins to maintain their stability. However,
   further studies are required to elucidate the detailed mechanisms
   underlying these interactions.

   The formidable presence of the BBB poses a substantial impediment to
   the direct administration of drugs for the treatment of gliomas,
   necessitating the identification of agents that can effectively
   modulate the TME and traverse the BBB. To address this, the CMap
   database was used to screen for readily available compounds capable of
   influencing the activity of miR-423-3p. Based on the concentration of
   miR-423-3p and IC50 values, 15d-PGJ[2], a PPARγ agonist derived from
   cyclopentenone prostaglandins, was selected as a candidate drug for
   further research. It inhibits tumor progression by affecting both tumor
   cells and the TME [[213]71]. Specifically, it suppresses CD133+ brain
   tumor stem cells by downregulating the EGF/bFGF and Tyk2–Stat3 pathways
   [[214]36]. Our analysis revealed that 15d-PGJ[2] potently inhibits the
   activation of NHAs induced by H-GDEs or miR-423-3p. In addition,
   15d-PGJ2 counteracted the enhanced proliferation of glioma cells
   following exposure to reactive astrocytes. Consistent with this study,
   previous studies have shown that primary and cyclopentenone
   prostaglandins, including 15d-PGJ[2], downregulate astrocyte
   proliferation [[215]72]. Moreover, 15d-PGJ[2] inhibits IL-6 expression
   in pituitary cells [[216]73], aligning with this study. These findings
   suggest that 15d-PGJ[2] is a promising therapeutic agent for inhibiting
   glioma progression, particularly by modulating reactive astrocytes and
   the TME.

   In summary, this study identified a positive feedback mechanism through
   which gliomas trigger astrocyte activation, which in turn promotes
   glioma malignancy. Furthermore, gliomas promoted the transformation of
   NHAs into RAs through the secretion of miR-423-3p-containing exosomes,
   which subsequently induced astrocytic reactivity through autophagy.
   This effect was inhibited by 15d-PGJ2. This study highlights the
   importance of the TME in brain tumor research, shifting the focus from
   only the tumor. Astrocytes and other cells in the TME may be a
   completely new research perspective, not only in glioma research but
   also in other brain research, such as neurodegenerative disorders.
   Additionally, this study provides new biological insights into the
   diagnosis and treatment of gliomas. Changes in the levels of exosomes
   and miR-423-3p may help in the technological improvement of liquid
   biopsies. The activation of reactive astrocytes can aid in the
   diagnosis of glioma. The inhibition of miR-423-3p and autophagy may
   disrupt the formation of the TME and ultimately impair glioma
   progression. The use of 15d-PGJ2 may help treat gliomas from a new
   perspective.

Statistical analysis

   All data were compared using one-way ANOVA, Student’s t-test et al. to
   detect differences using GraphPad Prism 8. Data from the experiment are
   shown as means ± SD. All tests were considered statistically
   significant at P < 0.05.

Supplementary information

   [217]Supplymental Material^ (31.7KB, docx)
   [218]Supplementary Figure 1^ (37MB, tif)
   [219]Supplementary Figure 2^ (38.1MB, tif)
   [220]Supplementary Figure 3^ (35.1MB, tif)
   [221]Supplementary Figure 4^ (41.9MB, tif)
   [222]Supplementary Figure 5^ (41.3MB, tif)
   [223]Supplementary Figure 6^ (42.7MB, tif)
   [224]Supplementary Figure 7^ (43MB, tif)
   [225]original western blot^ (22.1MB, docx)
   [226]reproducibility checklist^ (2.1MB, pdf)

Acknowledgements