Hypoxic Glioma Stem Cell–Derived Exosomes Containing Linc01060 Promote Progression of Glioma by Regulating the MZF1/c-Myc/HIF1α Axis

Gliomas are derived from the neuroectodermally derived glial cells, namely, astrocytes, oligodendrocytes, or ependymal cells and their multipotent progenitors. They account for 40% to 50% of the craniocerebral tumors and are the most common primary intracranial tumors (1). Its annual incidence rate is 3 to 8 people per 100,000 (2). The mechanism of this type of tumorigenesis is currently unclear, which often infiltrates into the normal brain tissue (NBT). Current treatments, such as surgery, radiotherapy, chemotherapy, and immunotherapy, cannot completely cure glioma, and have not greatly improved the clinical prognosis of patients (3). Therefore, we urgently need to explore the mechanism behind the occurrence and development of the gliomas and find potential markers for timely intervention.

Glioma stem cells (GSC) exert a vital role in tumor proliferation, survival, recurrence, and metastasis (4). In essence, they maintain an active tumor cell population via self-renewal and infinite proliferation. The movement and migration abilities of the GSCs make the transfer of tumor cells possible. The GSCs can remain dormant for long and be responsible for multidrug resistance (5). They are more resilient than the bulk tumor cells. Thus, the tumor often relapses after a period of time despite the success of the conventional tumor treatment in eliminating most of the tumor cells (4).

Exosomes are small, membrane-bound, extracellular vesicles that contain proteins and genetic materials. Under normal and pathologic conditions, a variety of cells secrete exosomes. They are mainly derived from the multivesicular bodies and released outside the cell after the fusion of the outer membrane of the multivesicular body and the cell membrane (6). When secreted from the host cell into the recipient cell, an exosome can regulate the biological activity of the recipient cell through the proteins, nucleic acids, lipids, and the other materials carried by it (7). Accumulating studies have revealed that exosomes act as a medium of communication between cells and participate in several cellular processes, such as antigen presentation, immune responses, differentiation, migration, invasion, and many more (8).

Long noncoding RNAs (lncRNA) refer to noncoding RNAs with a length of more than 200 nucleotides (9). lncRNAs have important regulatory function, and are involved in various biological processes and pathways. They are inseparably correlated with the occurrence and progression of various diseases, thus becoming a research hotspot in the past few years (10). The human genome produces a greater number of lncRNAs than the coding RNAs. The function of most lncRNAs is currently unknown barring a few exceptions. GSCs can secrete exosomes, especially under hypoxic conditions, and the exosomes contain a lot of lncRNAs. How lncRNAs in the hypoxic GSC (H-GSC)-derived exosomes promoting the progression of glioma is currently unclear. Thus, studying these in depth would be beneficial in understanding their roles and mechanisms in different diseases.

To clarify which lncRNA is encapsulated by exosomes derived from H-GSCs, we collected and isolated exosomes from the supernatants of glioblastomas (GBM), normoxic GSCs (N-GSC), and H-GSCs and performed high-throughput sequencing. From the sequencing data, long intergenic nonprotein-coding RNA 1060 (Linc01060) was one of the highest-scoring candidates among the upregulated molecules. Analysis of the clinical data showed that the expression levels of Linc01060 were inversely related to the prognosis of the patients. Functionally, Linc01060 promoted the progression of gliomas. In addition, Linc01060 was mainly secreted by H-GSCs, transported to tumor cells through exosomes to play a carcinogenic role, and was significantly upregulated under hypoxic conditions. Interestingly, HIF1α bound to the hormone response elements (HRE) in the Linc01060 promoter to stimulate its transcription. Mechanistically, Linc01060 directly bound with the myeloid zinc finger 1 (MZF1), one of the Kruppel family proteins that activates or represses gene transcription by binding to the promoters. Abnormal expression of MZF1 is documented in numerous tumors (11, 12). Herein, we observed that Linc01060 stabilized MZF1 from ubiquitination-mediated degradation, accelerated its nuclear translocation, and promoted MZF1-mediated c-Myc transcriptional activities. In addition, c-Myc enhanced the accumulation of HIF1α at posttranscriptional level. Meanwhile, HIF1α bound to the HREs in the Linc01060 promoter to enhance the transcription of Linc01060. The positive feedback loop formed by Linc01060, MZF1, c-Myc, and HIF1α triggers a cascading amplification for the development of glioma. This could be a potential therapeutic target against gliomas.

All fresh glioma specimens were obtained from the patients operated at the Wuhan Union Hospital (Wuhan, China). All specimens were in quadruplicates, each stored in (i) tissue preservation solution for primary culture, (ii) 4% paraformaldehyde for fixation, (iii) ISH preservation solution for fixation, and (iv) liquid nitrogen for subsequent experiments. Before collecting the specimens, we obtained “informed written consent” from patients. No patient received radiotherapy or chemotherapy before the surgery. This research was performed according to the International Ethical Guidelines for Biomedical Research Involving Human Subjects issued by the Council for International Organization of Medical Sciences. All patients were graded according to the 2016 edition of the World Health Organization classification criteria for central nervous system tumor.

Normal human astrocytes (NHA) and human astrocyte (HA) cells were obtained from the ScienCell Research Laboratories and cultured in the astrocyte medium. Human glioma cell lines [H4, LN-18, A-172, U-118MG, U-251MG, and U-87MG (GBM of unknown origin)] were obtained from the ATCC. All cell lines were authenticated by short tandem repeat and verified to be free of Mycoplasma contamination. For detailed cell culture protocol, please refer to our previous articles (13, 14).

The exosomes were purified by ultracentrifugation according to standard operating procedures. For more detailed protocols, please refer to our previous articles (13).

The U-87MG cells were transfected with the siRNAs targeting Linc01060, MZF1, HIF1α and c-Myc using Lipofectamine 3000 (Invitrogen). We constructed the Linc01060 pcDNA3.1 [1060-OE (overexpression)], MZF1 pcDNA3.1 (MZF1-OE), c-Myc pcDNA3.1 (c-Myc-OE), and HIF1α pcDNA3.1 (HIF1α-OE) vectors, inserting the respective genes into the pcDNA3.1 vector (Invitrogen). Empty pcDNA3.1 vector (vectors) was used as control. All the siRNAs and sh1060 and shNC lentiviral vectors were obtained from GeneChem Co., Ltd. Supplementary Table S1 lists the sequences of relative siRNAs and short hairpin RNAs (shRNA). For detailed protocol, please refer to our previous articles (15, 16).

The chromatin immunoprecipitation (ChIP) assays were conducted using ChIP assay kits (Upstate Biotechnology). The cells were cross-linked with formaldehyde followed by sonication. We pretreated the cell lysate with protein A/G beads before incubating with the beads coated with anti-HIF1α antibodies. IgG acted as negative control. We extracted DNA from the complexes using the DNA extraction kits (QIAGEN) and performed quantitative real-time PCR (qPCR). Supplementary Table S2 lists the primers used in ChIP-qPCR.

The dual-luciferase reporter assays were performed according to the manufacturer's protocols (Promega). Briefly, the luciferase reporter for detecting the transactivation of Linc01060 was constructed by annealing complementary oligonucleotides containing four HIF1α putative binding sites [wild type (WT) and mutant type (MUT)] and inserting into the pGL3-control firefly luciferase reporter gene vector. For more detailed protocol, please refer to our previous article (13).

The protocol was conducted as described previously (16–18). Supplementary Table S3 lists the antibodies used.

Total RNA was isolated from the cells, serum, exosomes, and cerebrospinal fluid using TRIzol and TRIzol LS reagents (Life Technologies). The miRNAs were reverse transcribed using the Mir-X miRNA First-Strand Synthesis Kit (Clontech). Real-time PCR was performed using the SYBR Green PCR Master Mix (Takara) and the primers listed in Supplementary Table S4. The mRNA levels were measured with the 7500 Fast Real-Time PCR Systems (Applied Biosystems). GAPDH was used as internal control.

Subcellular fractionations of RNAs were conducted using the PARIS Kit (Ambion). We further analyzed the RNA content in the cytoplasm and nucleus using qRT-PCR. β-actin and U6 served as internal references for cytoplasm and nucleus, respectively. We separated and extracted proteins from cytoplasm and nucleus using the Minute Cytoplasmic and Nuclear Extraction Kit (Invent Biotechnologies). α-Tubulin and lamin B1 were the internal references for cytoplasm and nucleus, respectively.

FISH assays were performed on the tissue sections with FISH-kits (Boster Bio) and the Linc01060 detection probes (RIBOBIO). The FISH, immunofluorescence (IF), and IHC assays and the scoring techniques were performed as described previously (13, 17, 18). The extent of positive staining was categorized as follows: no positive as 0, 0%–10% positive as 1, 10%–30% positive as 2, 30%–70% positive as 3, and 70%–100% positive as 4. The staining intensity was categorized as follows: no staining as 1, weak staining as 2, moderate staining as 3, and strong staining as 4. The staining index (SI) was calculated as the product of these two, and the cut-off values were SI = 8. Thus, samples with SI ≥ 8 had high expression, and samples with SI < 8 had low expression. Supplementary Table S3 lists the antibodies used in this experiment.

We tested the ability of cell proliferation of U-251MG and U-87MG using the Cell Counting Kit-8 (CCK-8) assay, 5-ethynyl-2′-deoxyuridine (EdU) cell proliferation assay, and colony formation assay. All experimental steps were performed according to the method described in our previous articles (15, 16).

We performed the coimmunoprecipitation (co-IP) experiment as reported earlier (14, 18). Supplementary Table S3 lists the details of the antibodies involved in this experiment.

Proximity ligation assay (PLA) was conducted with the Duolink In Situ Red Starter Kit Mouse/Rabbit (Sigma-Aldrich). Briefly, U-87MG cells were cultured on glass coverslips and fixed with 4% paraformaldehyde. Subsequently, the cells were permeabilized with 0.1% Triton X-100 and then blocked with blocking solution for 1 hour. Next, the cells were incubated overnight at 4°C with anti-MZF1 and anti-c-Myc antibodies (Abcam). Then, we added the positive and negative strand labeled PLA probes to bind to the corresponding primary antibodies. The cells were sequentially incubated with ligase for 0.5 hours and polymerase for 2 hours at room temperature. Subsequently, the coverslips were mounted on the slide using Duolink In Situ Mounting Medium with DAPI.

We investigated the intracellular glucose uptake of the cells using fluorodeoxyglucose F 18 (¹⁸F-FDG), as reported previously (19, 20). The lactic acid content was determined in the cell supernatant according to the manufacturer's protocols (Agilent Seahorse XF Analyzers). Next, we used reagents to lyse the cell pellet and measure ATP levels with Seahorse XF (Agilent).

The changes in cellular metabolism were monitored using the Seahorse Bioscience XF 24 Extracellular Flux Analyzer (Agilent). Cells transfected with si1060, control siRNAs, 1060-OE plasmid, and control plasmid were cultured overnight in XF 24-well plate (5 × 10⁴ cells/well) and then starved for serum for one day. Extracellular acidification rate (ECAR) assays were tested with the Seahorse XF Glycolysis Stress Test kits (Agilent Technologies) and oxygen consumption rate (OCR) assays were measured using the Seahorse XF Cell Mito Stress Test kits.

Biotin-RNA pull-down assays were conducted as described previously (21, 22). Briefly, the full-length Linc01060 sequences were amplified using PCR and then reversely transcribed. Cellular proteins were lysed using a lysis buffer. Then, the samples were incubated with streptavidin agarose beads to capture the biotin-labeled Linc01060 probes. The pull-down complexes were measured by mass spectrometry (MS) or Western blotting.

RNA immunoprecipitation (RIP) assays were conducted with Magna RNA-binding protein immunoprecipitation kit (Millipore). Briefly, the cells were collected and lysed with RIPA buffer. The cell lysates were incubated with RIP buffers containing magnetic beads conjugated to human anti-MZF1 or IgG antibodies. Then, the coprecipitated RNAs were measured using qRT-PCR. Total RNA and IgG controls were also measured to ensure that the detected RNA signal was associated with MZF1.

For in vivo brain orthotopic xenografts, 8-weeks-old male BALB/c nude mice were randomly divided into the corresponding group (n = 5). For modeling, stereotaxic instrument and microsyringe were used to implant tumor cells 3 mm deep into the mouse brains. Meanwhile, we employed an indwelling cerebral monitoring device for subsequent continuous intracranial intervention. We performed in vivo imaging of the mice with orthotopic tumor xenografts using the IVIS Lumina III in vivo imaging system (PerkinElmer). The Institutional Animal Care and Use Committee has approved these studies. For more detailed operation of specific animal experiments, please refer to our previous studies (13, 18).

All statistical analyses were conducted with SPSS 23.0 software (SPSS, Inc.) and GraphPad Prism version 8.0 (GraphPad Inc.). All data were presented as the mean ± SD. Unpaired/paired Student t test for two groups or one-way ANOVA + Dunnett for more than two groups, which was used to assess statistically significant data. The χ² test, Pearson correlation, and one-way analysis of variance were also performed. Cox regression analysis and log-rank test were used to determine survival difference and HR. P < 0.05 was considered to be statistically significant.

We isolated primary tumor cells and GSCs from the surgical specimens of human GBM and identified them using flow cytometry and sphere formation assay. Flow cytometry results showed that, compared with primary tumor cells, the GSCs highly expressed the CD44, CD73, CD105, and CD133 stem cell markers (Supplementary Fig. S1A). Sphere formation assay exhibited that the tumor stem cells were more likely to form larger spheres than tumor primary cells (Supplementary Fig. S1B). We cultured the GSCs under normal or hypoxic conditions forming normoxic (N-GSCs) and hypoxic (H-GSCs) cells, respectively (Supplementary Fig. S1C). Unexpectedly, when we collected the condition media of the GBM, N-GSCs, and H-GSCs to treat the U-87MG cells, we found that the medium of the GSCs significantly promoted the proliferation, migration, and invasion of U-87MG, especially that of the H-GSCs (Supplementary Fig. S1D). To identify the potential exosomes-associated lncRNAs in GSCs, we performed lncRNA arrays using the exosomes from GSCs with or without hypoxia and the corresponding GBM primary cells. Linc01060 was one of the highest-scoring noncoding RNAs in hierarchical clustering (Fig. 1A). Supplementary Table S5 listed the gene list. To verify the presence of exosomes, we extracted the exosomes from the media of the GSCs, and identified their morphology, size, and surface markers using transmission electron microscopy (TEM), nanoparticle tracking analysis (NTA), and Western blotting (WB; Fig. 1B). Meantime, we also performed a Western blotting to confirm purity of exosomal preparations utilize exosome negative marker Calnexin (Supplementary Fig. S2A). Simultaneously, we observed that hypoxia promoted the release of exosomes (Fig. 1C).

Next, we examined the existence of extracellular Linc01060. When we treated the GSCs with RNase A alone, Linc01060 level in the condition medium did not change significantly; but on treatment with a combination of RNase A and Triton X-100, the level in the medium significantly reduced (Fig. 1D). Simultaneously, the level of Linc01060 in the exosomes was similar to that in the condition medium (Fig. 1E). These results indicated that exosomes are the main carriers of extracellular Linc01060.

Next, we verified the content of Linc01060 in the GSCs using qRT-PCR. Compared with the primary tumor cells, the GSCs overexpressed Linc01060, especially under hypoxic conditions (Fig. 1F). Linc01060 was evidently upregulated in the six glioma cell lines compared with the HA and NHA cell lines (Supplementary Fig. S3A). Overexpression or knockdown of Linc01060 in the GSCs resulted in the upregulation or downregulation of exosomal Linc01060, respectively (Supplementary Fig. S3B).

The FISH analysis demonstrated that Linc01060 was mainly located in the cytoplasm (Fig. 1G). Next, we incubated PKH67-labeled U-87MG cells with Dil-labeled GSCs exosomes. The IF results displayed a colocalization of the PKH67 lipid dye and Dil fluorescence in the incubated U-87MG cells, indicating that the cells effectively absorbed the exosomes (Fig. 1H). These results suggested that Linc01060 is contained in GSC-secreted exosomes and could be transferred to U-87MG cells. We measured the levels of Linc01060 in the U-251MG cells incubated with exosomes derived from GSCs/vector and GSCs/1060-OE. Linc01060 was evidently upregulated in the cells incubated with GSCs/1060-OE exosomes (Supplementary Fig. S3C). As expected, the GSCs/1060-OE exosomes failed to induce Linc01060 in U-251MG on treatment of Annexin V (Supplementary Fig. S3D). These results indicated that GSC-secreted Linc01060 could be transferred to glioma cells through the exosomes.

To elucidate the biological function of exosome-mediated Linc01060, we infected GSCs with lentiviral vectors containing different siRNAs and two shRNAs to effectively knockdown the Linc01060 gene. We extensively used si1060#2 siRNA in our study as it demonstrated the strongest suppression of Linc01060 (Supplementary Fig. S3E). We verified this knockdown using EdU cell proliferation assay, colony formation experiments, transwell migration and invasion assays, and animal experiments. All these experimental results indicated that GSC-derived exosomes could accelerate the proliferation, migration, and invasion of tumor cells. However, with annexin V treatment or knockdown Linc01060, the exosomes failed to promote these processes (Fig. 2A).

To probe the in vivo effects of H-GSC exosome on tumor growth, we treated U-87MG/Luc cells with PBS. We coimplanted N-GSC–derived exosomes or H-GSC–derived exosomes in the nude mice. After implantation, we injected PBS, N-GSC–derived exosomes, or H-GSC–derived exosomes in situ every 3 days. Three weeks later, in vivo animal imaging results showed that the signal was stronger in the mice injected with H-GSC–derived exosomes (Supplementary Fig. S3F). IF staining showed that U-87MG/Luc treated with H-GSC–derived exosomes had higher expression of Ki-67 protein and resulted in more aggressive gliomas (Supplementary Fig. S3G). Consistent with the above findings, the survival time of the group of mice injected with H-GSC–derived exosomes was significantly shorter than that of mice in other groups (Supplementary Fig. S3H). In addition, we coimplanted U-87MG/Luc cells treated with PBS, GSCs/Linc01060 exosomes, GSCs/Linc01060 exosomes with Annexin V, or GSCs/shLinc01060#2 exosomes in nude mice. After implantation, we injected the mice with PBS, GSCs/Linc01060 exosomes, GSCs/Linc01060 exosomes with Annexin V, or GSCs/sh-Linc01060#2 exosomes every 3 days. Three weeks later, in vivo animal imaging results displayed that the signal was stronger in the group of mice with GSCs/Linc01060 exosomes (Fig. 2A). IF staining showed that U-87MG/Luc cells treated with GSCs/Linc01060 exosomes had higher expression of Ki-67 protein (Fig. 2B). Consistent with the above findings, the survival time of mice injected with GSCs/Linc01060 exosomes was significantly shorter than that of the mice in other groups (Fig. 2C).

Accumulating evidence indicates that cytoplasmic lncRNA play a regulatory role in translation or posttranslation processes. Our data and previous studies have shown that Linc01060 was predominantly located in the cytoplasm (23). Because Linc01060 mainly located in the cytoplasm, we screened for Linc01060-interacting proteins using RNA pull-down and MS assays. On the basis of our MS data (Supplementary Fig. S4A) and RNA pull-down (Fig. 3A), we focused on the RNA-binding transcription factor MZF1. The RIP assays demonstrated that Linc01060 specifically enriched in the MZF1-immunoprecipitated complexes (Fig. 3B). The proteins extracted from Linc01060 pull-down assays indicated that MZF1 specifically bound to the sense strand of Linc01060 but not the antisense strand (Fig. 3C). To verify the specific regions within Linc01060 that bound to MZF1, we created four truncations of Linc01060 according to the secondary structures predicted with the AnnoLnc databases. Then, RNA pull-down assays exhibited that MZF1-specific binding sequences were located at 281–481 nt long region of the Linc01060 gene. To identify the interacting domains of MZF1, we conducted RIP assays with a variety of Flag-tagged MZF1 truncations. The A2 domain (358–599) of MZF1 bound to Linc01060 (Supplementary Fig. S4B and S4C).

Next, we explored the molecular function of Linc01060–MZF1 interaction. We observed that alteration of Linc01060 levels regulated the MZF1 protein levels (Fig. 3D). Meanwhile, with Linc01060 overexpression, the exosomes could upregulate the MZF1 protein levels. However, with annexin V treatment or knockdown Linc01060, the exosomes failed to promote the MZF1 protein levels (Supplementary Fig. S2B). The upregulation of MZF1 protein could possibly be a result of promoting transcription, enhancing translation, or inhibiting proteasome-mediated degradation. Yet, there was no significant difference in MZF1 mRNA levels in the Linc01060 knockdown or overexpressing cells (Supplementary Fig. S4D and S4E). Therefore, we conjectured that Linc01060 might stabilize the MZF1 protein. As expected, we found that the proteasome inhibitor MG-132 could rescue the phenomenon of downregulation of Linc01060, leading to a decrease in MZF1 protein levels (Fig. 3E). Cycloheximide chase assay results displayed that MZF1 had a shorter half-life in cells downregulated for Linc01060; similarly, it had a longer half-life in cells overexpressing Linc01060, compared with the controls (Fig. 3F). This indicated that Linc01060 might regulate MZF1 by inhibiting its proteasome-mediated degradation.

The ubiquitin-proteasome system is a crucial pathway for protein degradation. Thus, we probed whether Linc01060 is related to the ubiquitin-mediated degradation of MZF1 with in vitro ubiquitination assays. The knockdown of Linc01060 enhanced the ubiquitination of the MZF1 protein in U-87MG and U-251MG cells. On the contrary, the ubiquitin-mediated degradation of MZF1 evidently diminished in the U-251MG and U-87MG cells with Linc01060 overexpression (Fig. 3G; Supplementary Figs. S2C and S5A). These results indicated that Linc01060 stabilized the MZF1 protein and protected it from ubiquitin-mediated proteasomal degradation.

Our IF data revealed that Linc01060 could regulate the total and nuclear levels of MZF1. In the U-87MG negative control cells, MZF1 mainly localized in the cytoplasm and a small amount present in the nucleus. In the Linc01060 knockdown cells, MZF1 was almost undetectable in the nucleus. However, in the Linc01060-overexpressed cells, MZF1 levels in the nucleus significantly increased (Fig. 4A). Analysis of the nuclear and cytoplasmic lysates from the U-251MG and U-87MG cells indicated an accumulation of MZF1 predominantly in the cytoplasm compared with the nucleus. In the U-87MG cells with Linc01060 silencing, the nuclear and cytoplasmic contents of MZF1 significantly reduced. However, ectopically expressed Linc01060 increased the MZF1 content in the nucleus and cytoplasm, consistent with our IF results (Fig. 4B).

As a bifunctional transcription factor, MZF1 inhibits or activates gene transcription by binding to a promoter. Previous studies have implicated MZF1 in cancer; it translocates into the nucleus and activates c-Myc, thus having a carcinogenic effect (11). In this study, MZF1 was a downstream target of Linc01060 and promoted tumor progression, prompting us to speculate the role of Linc01060 in the MZF1/c-Myc signal axis. First, we used co-IP experiments to verify whether Linc01060 could modulate the interaction between MZF1 and c-Myc. Expectedly, knocking down Linc01060 greatly weakened the interaction between MZF1 and c-Myc, and overexpressing Linc01060 significantly enhanced the function (Fig. 4C). In addition, we used the in situ PLA to visualize the natural protein complexes (Fig. 4D; Supplementary Fig. S6A). Under Linc01060-silencing conditions, there were fewer PLA-positive protein complexes, including MZF1 and c-Myc. In contrast, we observed high-density MZF1/c-Myc clusters in the Linc01060 overexpression groups, corresponding with the co-IP results. In addition, RNA pull-down and RIP results did not exhibit a direct interaction between the Linc01060 gene and c-Myc protein (Supplementary Fig. S6B and S6C).

Next, we probed whether Linc01060 could regulate MZF1-mediated c-Myc transcriptional activity. The luciferase reporter assays indicated that overexpression of MZF1 alone and its expression with Linc01060 could enhance the transcriptional activities. The relevant rescue experiments showed that the depletion of both Linc01060 and MZF1 could lead to a significant decrease in c-Myc transcriptional activity (Fig. 4E). Then, we examined the expression of the target genes of MZF1 and c-Myc involved in the Warburg effect, such as HK, PGK1, LDHA, GLUT1, and ENO1. Cotransfection of Linc01060 and MZF1 resulted in relatively high expression of all these genes; while, the simultaneous knockdown of Linc01060 and MZF1 led to their lower expression (Fig. 4F). In short, our results indicated that Linc01060 enhanced the MZF1-mediated c-Myc transcriptional activity.

We investigated the factors causing high expression of Linc01060 in gliomas. Previous studies have reported that c-Myc enhances the accumulation of HIF1α at the posttranscriptional level (24). Our results were consistent with these reports (Supplementary Fig. S6D and S6E). Besides, our results indicated that c-Myc stabilized the HIF1α protein, protecting it from ubiquitin-mediated proteasomal degradation (Fig. 5A; Supplementary Fig. S2C). JASPAR and UCSC analyses of the region approximately 2 kb upstream of the genome sequence of Linc01060 revealed one predicted HIF1α binding site in the promoter region (Fig. 5B; refs. 25, 26). Pretreatment with hypoxia or CoCl₂ for one day, evidently elevated the Linc01060 and HIF1α levels (Fig. 5C). In contrast, the knockdown of HIF1α gene under normoxic and hypoxic conditions significantly inhibited the transcription of Linc01060 (Fig. 5D and E). Subsequent ChIP-qPCR results demonstrated that HIF1α directly bound to the same chromatin fragment of the promoter region of the Linc01060 gene (Fig. 5F). To verify whether HIF1α activates Linc01060 transcription, we cloned the Linc01060 5′UTR fragment containing the WT or MUT HRE-binding sequences into the promoter regions of the pGL3-control plasmids. Hypoxia increased the luciferase intensity in cells with the WT promoter, while there was no significant difference in cells containing the MUT promoter. However, knocking down the HIF1α gene significantly suppressed the luciferase intensity containing the WT promoter (Fig. 5G and H). In conclusion, our data proved that Linc01060 is a direct transcriptional target of HIF1α.

Bioinformatics predictions indicated that Linc01060 might be involved in the glycolytic metabolism of the tumor cells (Supplementary Fig. S6F). Our experimental results showed that the Linc01060/MZF1/c-Myc/HIF1α axis could promote the expression of glycolytic genes, such as HK, PGK1, and LDHA. In addition, MZF1, c-Myc, and HIF1α also promote the transcription of glycolytic genes (12, 20, 21).

We examined the involvement of Linc01060 in aerobic glycolysis. We knocked down Linc01060 in the U-87MG cells and overexpressed it in the U-251MG cells (Supplementary Fig. S6G). Silencing Linc01060 increased the maximum respiration in the OCR experiments, while ECAR profile demonstrated that knocking down Linc01060 significantly reduced glycolysis, glycolysis reserve, and glycolysis capacity in the U-87MG cells (Fig. 6A and B). In contrast, overexpression of Linc01060 exhibited opposite results in the U-251MG cells (Fig. 6C and D), indicating that Linc01060 promotes glycolytic metabolism. Besides, when Linc01060 was downregulated, the ¹⁸F-FDG uptake, lactate production, and cellular ATP levels significantly reduced in the U-87MG cells, while its overexpression in the U-251MG cells exhibited opposite results (Fig. 6E–G).

In addition, we measured the expression levels of several proteins involved in glycolytic metabolism. The results displayed that changes in Linc01060 expression could significantly affect the expression levels of these proteins (Supplementary Fig. S6H). All these data indicated that Linc01060 could regulate glycolysis metabolism in gliomas.

We studied the role of the MZF1/c-Myc/HIF1α signal axis in context of the biological function of Linc01060 with relevant rescue experiments. We transfected the MZF1, c-Myc, or HIF1α plasmids into the U-87MG cells with downregulated Linc01060. These significantly rescued the inhibition of cell proliferation (Fig. 7A and 7B; Supplementary Fig. S7A and 7B), glucose uptake (Fig. 7C; Supplementary Fig. S7C), lactate production (Fig. 7D; Supplementary Fig. S7D), and cellular ATP levels (Fig. 7E; Supplementary Fig. S8A). Using WB and qRT-PCR, we also verified the efficiency of MZF1/HIF1α knockdown or overexpression (Supplementary Fig. S8B). We transfected the MZF1, c-Myc, or HIF1α siRNAs into the U-251MG cells with overexpressed Linc01060. These consequently reduced the Linc01060-induced increase in cell growth (Supplementary Figs. S7E–S7F and S8C), glucose uptake (Fig. 7F; Supplementary Fig. S8D), lactate production (Fig. 7G; Supplementary Fig. S8E), and cellular ATP levels (Fig. 7H; Supplementary Fig. S8F). Of course, we also validated the results of Fig. 7 in condition of treatment of Linc01060-containing exosomes (Supplementary Fig. S9A–S9H). All these results proved that Linc01060 depends on the MZF1/c-Myc/HIF1α signal axis to regulate the cell proliferation and aerobic glycolysis.

To investigate whether the increased levels of exosomal Linc01060 in serum or cerebrospinal fluid (CSF) are related to glioma development, we isolated circulating exosome from the serum and CSF of patients with glioma and healthy people. Meantime, we also characterized exosomes from serum by NTA analysis, TEM and Western blotting from five cases (Supplementary Fig. S2D). The qRT-PCR results demonstrated that the patients with tumor had higher levels of Linc01060 in the circulating exosomes, especially in the high-grade glioma tissues, compared with the healthy controls (Supplementary Fig. S10A). Also, we tested the expression levels of Linc01060 in the circulating exosomes of patients with glioma before and after the operation. As expected, 80% (20/25) of the patients exhibited a dramatic decline in the level of Linc01060 in the circulating exosomes after removal of glioma tissues (Supplementary Fig. S10B).

We measured the Linc01060/MZF1/c-Myc/HIF1α expression levels in NBT, low-grade glioma tissues (LGG), and high-grade glioma tissues (HGG). The tumor tissues displayed higher expression of Linc01060/MZF1/c-Myc/HIF1α, especially in the HGG, compared with NBT (Supplementary Fig. S11A–S11D). Besides, the Linc01060/MZF1/c-Myc/HIF1α signal axis was directly proportional to the expression of each other (Supplementary Fig. S11E–S11G and Supplementary Fig. S10C–S10E). Furthermore, Linc01060 expression level in the circulating exosomes from serum and CSF positively associated with that in the glioma tissues (Supplementary Fig. S11H and S11I). This indicated that the upregulation of Linc01060 in glioma tissues could result in the elevated Linc01060 levels in the circulating exosomes. In addition, FISH, IF, and IHC results exhibited that Linc01060 levels positively correlated with the MZF1/c-Myc/HIF1α signal axis (Supplementary Fig. S11J and S10F). Therefore, our clinical data strongly suggested that a high level of Linc01060 in the circulating exosomes is related to glioma progression.

We also assessed and contrasted the Linc01060 levels with different clinicopathologic characteristics. The Linc01060 levels significantly correlated with the tumor size (P = 0.003), Karnofsky performance scale (KPS; P = 0.006), tumor grade (P = 0.017), and tumor recurrence (P < 0.001; Supplementary Table S6). Our previous research demonstrated that KPS could be used as an indicator of glioma, confirming the results of this study (17, 27). The univariate and multivariate regression analyses showed that Linc01060 level was an independent prognostic indicator for patients with glioma with significant HRs [HR, 2.536; 95% confidence interval (CI), 1.264–4.372; P = 0.018; Supplementary Table S7]. Collectively, our data indicated that Linc01060 might be a potential biomarker for glioma.

The interaction between tumor and stromal cells is critical for tumorigenesis. Here, we demonstrated that H-GSCs released exosomes that were internalized by the glioma cells, resulting in a significant increase in the tumor cell proliferation, glycolytic metabolism, and glioma tumorigenicity. Besides, we showed that H-GSC–derived exosomes contained highly expressed and enriched lncRNAs. Linc01060 was the most abundant lncRNA in these exosomes and significantly promoted tumor cell proliferation and glycolytic metabolism in vitro. The delivery of Linc01060 by the H-GSC–derived exosomes resulted in the upregulation of the MZF1/c-Myc/HIF1α signal axis in the recipient glioma cells (Supplementary Fig. S12).

Although several studies have reported the interplay between the stromal cells and TSCs (28, 29), only a few explored the interactions of H-GSC–derived exosomes with tumor cells. This connection between the tumor and stromal cells is an untapped field in cancer biology. Previous studies have indicated that stem cell–derived exosomes can promote tumor progression in numerous types of tumors, thereby supporting the results of our study (30, 31). Recently, researchers discovered that astrocyte-derived exosomes carrying miRNA promote brain metastasis (32) and GBM-derived exosomes transfer RNA to the microglia/macrophages in the brain (33). Our data indicated that H-GSC–derived exosomes not only promoted the growth of glioma, but also enhanced their glycolytic metabolism, leading to increased tumor burden and reduced survival rate in vivo.

Previous research demonstrated that Linc01060 suppressed the progression of pancreatic cancer by regulating the vinculin–focal adhesion axis and was predominantly located in the cytoplasm (23), which was consistent with our results. Furthermore, growing evidence shows that lncRNAs bind to specific proteins to perform their biological functions (34). The regulatory role of the lncRNAs in their interaction with proteins involves numerous aspects, such as regulating the expression and activity of the protein, changing the subcellular localization of the protein, or acting as a structural component (35). When we probed the proteins that interacted with Linc01060 using the RNA pull-down assays, the MZF1 protein, a member of the Kruppel family, drew our interest, as it is essential for the proliferation, migration, and differentiation of hematopoietic cell (36, 37). As a bifunctional transcription factor, MZF1 consists of 13 zinc finger domains and inhibits or activates gene transcription by binding to the promoter region of a gene (37). MZF1 is vital in tumorigenesis and cell invasion. Abnormally expressed MZF1 induces malignant transformation of the NIH3T3 cells and promotes tumor formation in athymic mice (38). Besides, MZF1 could promote aerobic glycolysis and neuroblastoma progression (12). MZF1 is involved in the progression of many solid tumors, such as cervical cancer (39), hepatocellular carcinoma (40), colorectal cancer (41), breast cancer (42), and lung cancer (11). Herein, our data demonstrated that Linc01060 physically bound to MZF1 and mediated the stability of MZF1 protein, preventing it from the ubiquitin-mediated proteasomal degradation. Besides, our results suggested that Linc01060 promoted the nuclear translocation of MZF1 (Fig. 4).

Overexpression of MZF1 resulted in the transactivation of the anexelekto promoter and enhanced the ability of colorectal cancer cells to migrate, invade, and metastasize (41). MZF1 promoted the migration and invasion of liver tumor cells by enhancing the transcription of protein kinase Cα (PKCα; ref. 40). Our previous research revealed that PKCα is involved in the progression of glioma (17). In lung adenocarcinoma, MZF1 enhanced the transcription of c-Myc, thus promoting tumor cell proliferation, migration, and invasion (11). In neuroblastoma, MZF1 promoted the growth and invasion of tumor cells by enhancing aerobic glycolysis, thus inducing the progression of neuroblastoma (12). Besides, MZF1 promoted the transcription of HK2 and PGK1 genes, which are key for aerobic glycolysis (12). MZF1 inhibited the migration and invasion of cervical cancer cells by suppressing the transcription of the matrix metalloproteinase-2 (39). These results indicate that MZF1 exerts carcinogenic or tumor suppressive effects by modulating the transcription of genes related to tumor progression. Herein, our data showed that MZF1 could promote the transcription of HK2, PGK1, and c-Myc genes, enhancing aerobic glycolysis and, in turn, inducing tumor progression. This suggests an oncogenic role of MZF1 in glioma. Meanwhile, MZF1 and c-Myc can directly bind, and Linc01060 affected the efficiency of this combination (Fig. 4).

A proto-oncogene and a crucial regulator, c-Myc is involved in a series of cellular processes, such as proliferation, differentiation, and apoptosis (43). It is highly expressed in various human cancers (20, 44, 45). Meanwhile, accumulating studies have reported that c-Myc was highly expressed in gliomas and involved in the progression of gliomas (46–48). Studies have reported the role of c-Myc in the emergence of radiation resistance against tumors (49). For example, targeting c-Myc enhanced the radiosensitivity of the embryonic rhabdomyosarcoma cells by promoting DNA damage and radiation-induced apoptosis and inhibiting the expression of DNA repair proteins (50). In addition, the upregulation of c-Myc in PKH26⁺ cells inhibited DNA repair and promoted radiation resistance by modulating the CHK1/2 inhibitor (51). Furthermore, the downregulation of c-Myc in HeLa cells significantly damaged the DNA double-strand break repair ability and enhanced the sensitivity to ionizing radiation by suppressing radiation-induced ATM phosphorylation and DNA-PKcs activity (52). c-Myc regulates aerobic glycolysis as a key oncogene in the metabolic reprogramming, making it a vital switch in the tumor cells to turn on/off their metabolic activities (53–55). A study reported that c-Myc could promote the transcription of LDHA (20). In this study, we found that c-Myc was upregulated in gliomas, and it enhanced the accumulation of HIF1α at posttranscriptional level, consistent with previous reports (24). Besides, our data revealed that c-Myc stabilizes HIF1α protein, protecting it from the ubiquitin-mediated proteasomal degradation (Fig. 5A).

An increasing number of studies have indicated that HIF1α is highly expressed in many solid tumors. The growth of these tumors is significantly restricted after knocking out HIF1α, indicating that HIF1α has a protective effect on hypoxic cells (56, 57). Our previous research also showed that HIF1α was upregulated in gliomas and promoted tumor progression (27). Recently, studies have reported the key role of lncRNAs in hypoxia-related energy metabolism (58, 59). For example, hypoxia-inducible lncRNA, LncHIFCAR, directly bound to HIF1α, thereby promoting the recruitment of p300 cofactor. HIF1α then bound to the promoter region of the target gene, thereby promoting oral cancer progression (22). In addition, lincRNA-p21, an lncRNA induced by hypoxia, could in turn modulate the stability of HIF1α. This positive feedback loop under hypoxic conditions promoted the glycolysis process (60). Our data indicated that Linc01060 promoted metabolic plasticity and enhancing a proliferation advantage under hypoxial stress. Moreover, the expression levels of Linc01060 rose under hypoxia. Mechanistic research proved that HIF1α could promote the transcription of Linc01060.

Numerous studies have emphasized the emerging role of exosomes as biomarkers in cancer diagnosis and prognosis assessment. Exosomal macrophage migration inhibitory factors of the pancreatic ductal adenocarcinoma cells (PDAC) could be conducive to the formation of the premetastatic niche in the liver and might be used to diagnose PDAC liver metastases (61). Exosomal integrin is a potential biomarker for the prediction of organ-specific metastasis (62). It is worth noting that the exosomal miR-25-3p from liposarcoma promoted its progression by forming a supportive microenvironment. It might be used as biomarker to predict the prognosis and therapeutic effects against liposarcoma (63). Besides, the miR-25-3p levels in the circulating exosomes may help in timely intervention by diagnosing the colorectal cancer metastasis and screening the patients at high risk for metastasis (64). In addition, our previous research proved that miR-182-5p level in the circulating exosomes from plasma or CSF is a potential biomarker for patients with glioma (13, 65). In this study, the Linc01060 level in the circulating exosomes from patients with HGG was higher compared with those from patients with LGG, especially from the healthy donors. In addition, the Linc01060 level dropped dramatically in most of the patients with glioma with resection surgery. Notably, the level of Linc01060 in the circulating exosomes from serum or CSF positively associated with that in the tumor tissues. In addition, the experimental design also has shortcomings. In the future, our research team must correct this bad habit. We only used one for experiments though we tested two siRNAs for Linc01060. Certainly, it is more stringent to use both analyses to limit the nonspecific effects of siRNA.

Collectively, our clinical data proved that Linc01060 level in the circulating exosomes from serum or CSF is a potential biomarker for monitoring progression of glioma.

No disclosures were reported.

J. Li: Conceptualization, data curation, software, formal analysis, methodology, writing-original draft, writing-review and editing. T. Liao: Data curation, software, methodology. H. Liu: Resources, software, methodology. H. Yuan: Resources, data curation, software, validation. T. Ouyang: Conceptualization, resources, software, formal analysis. J. Wang: Data curation, methodology. S. Chai: Data curation, methodology. J. Li: Conceptualization, resources, supervision. J. Chen: Resources. X. Li: Conceptualization, resources, supervision. H. Zhao: Conceptualization, resources, supervision. N. Xiong: Conceptualization, resources, supervision, funding acquisition, writing-review and editing.

Thanks to all technical staff of Wuhan Cell Learning Technology Co., Ltd for their technical support to the cultivation of primary cells and stem cells in this experiment. This study was supported by NSFC grants (no. 81671210, no. 30801180, no. 81371380) and Study on the Countermeasures for Hubei to Build a Medical Industry Innovation Consortium (no. 20001352).

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.