Novel mutations of SRPX facilitate the stemness and malignant progression of glioma

Abstract

Background

Explorations of genomic profiles are of great clinical significance for glioma due to the high tumor heterogeneity and stemness. High-depth sequencing enabled to identify driver genes and potential treatment targets in glioma.

Methods

A whole exome sequencing analysis of 27 Chinese patients was conducted and somatic mutation signatures were analyzed using validated computational methods. The biological roles of SRPX mutations and subsequent regulatory mechanisms were revealed by common experimental methods.

Results

We intriguingly found that SRPX drove glioma progression with two frequent in-frame deletion mutations of p.L14DEL and p.L23DEL. Mechanistically, both the two mutations of SRPX promoted binding with transcription factor AHR on its promoter, upregulated gene transcription of itself, then activated EGFR/ Akt/ Nestin pathway and contributed to aggressive tumor phenotypes and animal tumor growth. Further, knockdown of AHR or application of Akt inhibitor suppressed the oncogenic role of mutated SRPX.

Conclusions

Our study highlighted the landscape of glioma in revealing a non-distinctive mutation and signaling pathway profile between low and high grades. More importantly, we identified a novel role of SRPX mutations in acceleration to stemness and malignant progression, which could provide new targets in improving outcomes of glioma.

Introduction

As one of the most common and lethal primary tumours of the central nerves system, glioma has been classified into grade I to IV based on histological malignancy. Grade I to II are considered to be low grade glioma (LGG) and grade III to IV belong to high grade glioma (HGG) [1]. Comparing to LGG, HGG take account for ~80% and have extremely poorer outcomes and higher mortality in adult glioma with a median survival of 12–15 months and a 5-year overall survival less than 5% [2]. Despite of the comprehensive treatments of surgery, radio-, chemo-, immuno- and even cytokine-therapy, glioma still progresses aggressively and recurs frequently due to the tumour heterogeneity and stemness [3]. Heterogeneity of glioma exists in different grades, histologies, microenvironments, intra- and inter-tumour characteristics, which brought tremendous difficulties in understanding of aetiology, therapeutic effects, and long-term prognosis [4]. Therefore, explorations of tumour heterogeneity in molecular features are of great clinical significance for glioma.

With the rapid development of deep and high throughput sequencing, a general molecular profile of glioma has been established based on whole exome sequencing (WES) and whole genome sequencing (WGS). A set of genetic and epigenetic alterations has been uncovered to characterise glioma heterogeneity as they play a significant role in glioma initiation and progression during this decade. For instance, the first large scale genomic and transcriptomic analysis of 206 glioblastoma multiforme (WHO grade IV) conducted by The Cancer Genome Atlas Research Network (TCGA) in 2008 identified various critical somatic mutations including ERBB2, NF1, TP53, and PIK3R1, and revealed a link between O⁶methylguanine-DNA methyltransferase (MGMT) methylation and DNA mismatch repair pathway [5]. Besides of MGMT methylation, IDH1/2 mutation, 1p/19q co-deletion, BRAF V600E mutation and Histone H3 K27M mutation, ATRX loss of function and TERT promoter mutation also frequently occurred in glioma and some of which had already been introduced in diagnosis, guidance of precise therapies and prediction of clinical outcomes [6–9]. Hence, identifying additional driver genes and altered pathways in glioma offers the prospect of developing new targets and biomarkers in treatment of glioma.

SRPX, known as Sushi Repeat Containing Protein X-Linked gene and encoding protein as component of X chromosome, involves in autophagy, phagolysosome assembly, phagocytosis during disk shedding, cell adhesion and signal transduction [10, 11]. However, the function and mechanism of SRPX still remain uncertain in tumour. SRPX was firstly described as DRS gene in 1998 as it was isolated as a suppressor of viral oncogenes v-src transformation [12]. Early reports found that SRPX induced apoptosis via an ER-mediated pathway and suggested this pathway might contribute to the suppression of tumour formation [13]. Down-regulation of SRPX was detected in multiple myeloma and might functionally correlate to the over-angiogenic phenotype of multiple myeloma endothelial cells in active disease [14]. On the contrary, recent researches indicated that SRPX was highly expressed in cancer‑associated fibroblasts originated from ovarian carcinoma and promoted the tumour invasiveness through regulation of Ras homology family member A (RhoA) signalling pathway [15]. Head and neck squamous cell carcinoma derived exosomes increased SRPX in normal recipient cells and conferred cancer-associated pathologies [16]. Besides, SRPX were also found to upregulate in primary cultured invasive phenotype hepatocellular carcinoma cells [17]. Taken together, a diverse role of SRPX in malignancies requires further validation.

To furtherly search for potential driver genes and involved pathways, we undertook WES analysis of 27 Chinese gliomas and found two novel recurrent mutations of SRPX gene that might play an important role in glioma progression with a relatively high mutation frequency and conserved mutated sites in this cohort. Our study aims to exemplify the genomic profile of LGG and HGG glioma and elucidate the driving role and downstream pathways of SRPX in glioma, which ensures a further understanding of glioma onset and the potential to develop new targets for clinical strategies.

Materials and methods

Patient samples and clinical data

A total of 27 patients were enroled to this study. Soft tissue and matched peripheral blood samples were obtained under approval by Ethics Committee of Xiangya Hospital, Central South University. Histological diagnosis was centrally reviewed and validated according to WHO guidelines. All patients underwent surgery (25 with gross total resection, 2 with subtotal resection) and clinical data were summarised in Table S1.

Whole exome sequencing (WES)

WES and analysis were performed at the Genomics Laboratory of GenomicCare Biotechnology (Shanghai, China). DNA was extracted from thawed soft tissue and blood samples, then was sheared by sonicator to capture exome DNA. Exome DNA was sequenced on a NovaSeq-6000 sequencer (Illumina, San Diego, CA) to generate 150 bp paired-end reads. Image analysis and base calling were done using onboard RT3 software. After removing adaptors and low-quality reads, the reads were aligned to NCBI human genome reference assembly hg19 using the Burrows-Wheeler Aligner alignment algorithm and further realigned using the Genome Analysis Toolkit. Somatic single-nucleotide variants (SNV), indeletion (Indel), and copy number variations (CNV) were determined by the MuTect/ANNOVAR/dbNSFP31, VarscanIndel, and CNVnator softwares, respectively.

Bioinformatics analysis

The Gene Expression Omnibus (GEO) and CGGA database were used to compare the mRNA expression of SRPX in normal brain tissue (n = 20) and glioma tissue including LGG and GBM (total n = 325). CGGA were used to analyse the correlation of SRPX expression and patient overall survival. Mutation frequency distribution of SRPX was established across 25 projects in TCGA programme. Processing of the data was based on the guidelines of listed databases. Prediction of transcriptional factors of SRPX was performed based on four acknowledged TF database including PROMO, ChIP-Atlas, GTRD and hTFtarget.

Plasmids, siRNA and reagents

Plasmids of SRPX WT, SRPX L14 DEL, SRPX L23 DEL, SIX2 WT, SIX2 p.N285H, NOD1 WT and NOD1 p.N716H were constructed in pCDH-CMV-MCS-EF1-GFP vector and were all customed from YouBio (Changsha, China). Each plasmid construct was two-way sequenced to ensure the mutation site and type are exactly the same with WES data.

The siRNA of SRPX was purchased from RiboBio (Guangzhou, China) and siRNA of AHR, CEBPA, YY1 and JUN were purchased from Tsingke Biotechnology (Beijing, China). The sequence of siRNA was listed in Table S2. Transient transfection was performed using Lipofectamine 3000 reagent (Invitrogen, Carlsbad, CA) for plasmids and small interfering RNAs according to the manufacturer’s instructions.

Proteasome inhibitor MG-132 (HY-13259) and Temozolomide (HY-17364) were purchased from MedChemExpress (Monmouth Junction, NJ), Akt inhibitor (MK-2206) was purchased from Topscience (Shanghai, China).

Cell culture

LGG cell HS683 and HGG cell LN229 were obtained from Advanced Research Center of Central South University. All of the cell lines were cultured in DMEM (Gibco BRL, Grand Island, NY) supplemented with 10% FBS (Gibco BRL) at 37 °C in 5% CO_2.

Four immortalised glioma cell lines LN229 SRPX VEC/ WT/ p.L14DEL /p.L23 DEL were constructed using lentiviral infection technology. Plasmid of vector, SRPX WT, SRPX p.L14 DEL and SRPX p.L23 DEL were packed by packaging plasmids (Gag-pol, pRSV-Rev, and pCMV-VSV-G, #14887, #12253 and #8454, Addgene, respectively) under the co-transfection in 293 T cells. Glioma cells were infected by lentivirus for 48 h and screened with puromycin hydrochloride (60210ES25, Yeasen, Shanghai, China). Lentivirus was obtained 48 h and 72 h after co-transfection with Lipofectamine 3000 reagent (L3000015, Invitrogen) according to the instructions.

Cell viability assay

Cells were seeded in 96-well plates (3 × 10³ cells per well). Cell viability was detected by CCK8 kit (C0005, Targetmol, Shanghai, China) and incubated for 1 h at different times (0, 24, 48, 72 h). Absorbance was measured at 450 nm. All other viabilities were normalised to these values.

Sphere formation assay

Cells were seeded (2 × 10⁴ cells per well) to ultralow-attachment 6-well plates (Corning, NY) with serum-free DMEM/F12 medium containing EGF (PeproTech, NJ), basic Fibroblast Growth Factor (bFGF, Sigma-Aldrich, St. Louis, MO), B27 Supplement and N-2 Plus Media Supplement (Life Technologies, Gaithersburg, MD). Sphere were counted and photographed under a conventional microscope (DMI3000B, Leica) after incubation for 3 and 9 days.

Wound healing assay

Cells were plated in 6-well plates (2 × 10⁵ cells per well) and treated with different interferences including plasmid transfection, siRNA transfection, Akt or proteasome inhibitors for 48 h. When the cells reached a density of more than 90%, a crossing scratch was completed in the bottom of the plate using a pipette and then cells were replenished with fresh medium and photos were taken at 0, 24 h after scratching. Quantification of the wound closure was processed by ImageJ software.

Colony formation assay

Cells were seeded (1 × 10³ cells per well) in 6-well culture plates, treated with DMSO or  temozolomide (TMZ, 200 μM), and cultured for 10 days. Next, cells were fixed by 4% paraformaldehyde (BL539A, BioSharp Hefei, China) for 15 min and stained by crystal violet (V5265, Sigma-Aldrich) for 15 min. After staining, cells were photographed under a conventional microscope (DMI3000B, Leica) and counted manually.

Western blotting

Whole cell pellets were lysed at 4 °C for 30 min in IP buffer (87787, Thermo Fisher Scientific, Waltham, MA) and the protein concentration was determined by BCA protein assay (AR0197, Boster Biological Technology, Wuhan, China). Proteins were then separated by SDS PAGE and transferred to a PVDF membrane (Merck Millipore, Billerica, MA). The membranes were probed with primary antibodies, followed by a secondary antibody and antibodies used in this study were listed in Table S3. Protein bands were detected by the enhanced chemiluminescence reaction solution (36028-A, Yeasen).

Quantitative PCR (qPCR)

Extraction of total cell RNA was performed using TRIzol^TM (15596026, Thermo Fisher Scientific), and the cDNA was further obtained with a reverse transcription kit (K1621, Thermo Fisher Scientific) according to the manufacturer’s instructions. qPCR was performed with the ABI 7500 instrument (Foster City, CA) and SYBR Green (4309155, Life Technologies). Primer sequence used in the study were listed in Table S4.

Immunofluorescence (IF)

After attached to the poly-L-lysine (PLL) slides, adherent cells were fixed with 4% paraformaldehyde for 20 min, permeated with 0.5% Triton 100 for 15 min and blocked with 10% donkey serum for 1 h. The primary antibody anti-SRPX (A1217, 1:1000) and anti-p-EGFR (sc-81488, 1:1 000) were incubated overnight at 4 °C. Then, a fluorescent secondary antibody (SAB4600234 anti-rabbit, SAB4600042 anti-mouse, Sigma-Aldrich) was incubated for 45 min at room temperature, followed by DAPI staining solution (E607303, BBI Life Sciences, Shanghai, China). The fluorescence images were captured by a confocal microscope (LSM 510 META, Carl Zeiss, Germany).

Immunoprecipitation (IP)

Cells were lysed in IP buffer containing inhibitor cocktail (4693116001, Roche, Basel, Switzerland). The cell lysates were incubated with protein-G magnetic beads (B23201, Bimake, Shanghai, China) and antibodies anti-p-EGFR (AP0820, 1:200) at 4 °C overnight. Then, beads were washed with lysis buffer and the IP samples were analysed via immunoblotting.

Chromatin immunoprecipitation assay (ChIP)

The ChIP assay was performed using the EZ-Magna ChIP A-Chromatin IP Kit (17-408, Merck Millipore). LN229 cells were cross-linked with 1% formaldehyde, stopped by glycine, cleaned in ice-cold PBS, and collected for lysis. After lysis, sonication of cross-linked chromatin was performed to get chromatin fragments around 150–200 bp. Diluted chromatin (1%) was collected and used as input. Subsequently, anti-AHR and normal anti-Rabbit IgG (negative control) were added to incubate overnight at 4 °C for immunoprecipitating the chromatin. Protein/antibody/chromatin complexes were pelleted through the magnetic beads (B23201, Bimake, Shanghai, China). When the complexes were eluted by resuspension, proteinase K was added into and incubated at 62 °C for 2 h and 95 °C for 10 min to obtain separate DNA. The purified DNA was amplified by PCR and separated by agarose gel electrophoresis. Enrichment of the DNA fragment was detected by qPCR with SYBR Green I Master Mix (4309155, Life Technologies). The specific primers used were listed in Table S4.

Animal experiments

5-week-old female BALB/c athymic nude mice were randomly divided into 3 groups, each group contains 6 mice and all of mice were housed under pathogen-free conditions. SRPX WT/L14 DEL/L23 DEL LN229 cells (5 × 10⁶ cells in 100 μL PBS) were injected subcutaneously into the right flank. The tumour volume was measured using a vernier caliper every 3 days and tumour volume was calculated as below: V (mm³)  =  width² × length /2. When the tumour volume reached 800–1000 mm³ or mice were fed for more than 4 weeks, animals were euthanized and tumour tissue was removed. Each tumour tissue was segmented into two parts, one was lysed by RIPA buffer to detect protein expression via immunoblotting, the other was fixed with 10% buffered formalin and embedded in paraffin for the immunohistochemical experiments.

Immunohistochemistry (IHC)

27 human glioma samples of the patient enrolled in WES analysis were used to detect the expression of SRPX, p-EGFR, EGFR, p-Akt, Akt and Nestin. The clinical tissue and animal tissue sections were employed in two-step test kit (PV-9000, ZSGB-BIO, Beijing, China) and DAB kit (ZLI-9017, ZSGB-BIO), then all immunostained sections were re-stained with Hematoxylin (E607317, SangonBio, Shanghai, China). Photographs were taken at 400 magnification under a conventional microscope (DMI3000B, Leica) and processed using professional software. Information of primary antibodies was described in Table S3. The expression of each protein in IHC was semi-quantitatively evaluated using the method previously described [18]. Score of Low expression includes 0,1,2, Mid (middle) expression includes 3,4,6, and High expression includes 8,9,12, respectively.

Statistical analysis

All the statistical data are presented as mean ± S.E.M., statistical significance of the differences was determined using the one-way or two-way ANOVA Test. P <  0.05 was considered statistically significant.

Results

WES reveals mutation burden and copy number variations signatures of glioma

To figure out the mutation load and spectrum and explore potential driver genes in glioma in terms of tumour grade and recurrency, WES was performed on paired tumour and normal native genomic DNA in 27 patients consisting of four groups including low grade-primary (LGG-P, 9 cases), low grade-recurrent (LGG-R, 2 cases), high grade-primary (HGG-P, 13 cases), and high grade-recurrent (HGG-R, 3 cases). The age of diagnosis in our study ranged from 20 to 63 years old and the median-age-of-diagnosis was 44 years old (Table S1).

Overall, 10–40 critical mutated genes and 15–56 missense mutations were identified in each sample (Fig. 1a). No significant differences were found in the number of mutated genes or the total number of mutations between LGG (including LGG-P & LGG-R) and HGG (including HGG-P & HGG-R) groups, as well as in primary (including LGG-P & HGG-P) and recurrent (including LGG-R & HGG-R) tumour groups (Fig. 1b). The missense and non-missense mutation rate was also random among the LGG-P, LGG-R, HGG-P and HGG-R groups (Fig. 1c). That is to say, mutation burden of glioma is not certainly related to the grade or recurrency of the tumour. However, compared with LGG patients, HGG individuals presented higher CNV rates, suggesting a positive correlation of CNV alterations with malignancy of glioma. The mutation spectrum in our study was characterised by a predominance of C > T and T > G transitions, with no significant distinctions of base transition and transvertion in the four groups (Fig. 1d), which parallels to most of other glioma cohorts. Chromosomal gain-and-loss plots made by GISTIC indicated region of copy number alterations in arm of chromosome and showed that deletions were fairly typical to happen on chromosomes 1, 5 and 10, while amplifications frequently occurred on chromosomes 7 and 12 in glioma (Fig. 1e). No apparent tendency of copy number alterations was observed between LGG and HGG groups in chromosome. Since some previous WES researches indicated that age [19] and tumour location [20, 21] would exert different effects on mutation profile of glioma, here we also analysed the mutation burden including mutated genes and all kinds of mutations in this cohort but found no apparent correlation of mutation burden with age (Fig. S1A) or tumour locations (Fig. S1B).

Fig. 1. Mutation burden and CNV alteration spectrum are revealed by WES in glioma.

a Mutation burden including number of mutant genes and number of non-synonymous mutations in 4 groups: LGG-P, LGG-R, HGG-P and HGG-R. b Comparison of mutation burden in LGG/ HGG groups (left panel) and primary/ recurrent groups (right panel). LGG groups include LGG-P and LGG-R, HGG groups includes HGG-P and HGG-R. Primary group includes LGG-P and LGG-R, Recurrent group includes LGG-R and HGG-R. c Missense/ nonsense mutations and CNV gain/ loss spectrum of glioma patients in this study and patients ID were listed in 4 groups. d Base change spectrum of 4 groups, Ti transition, Tv transvertion. e Chromosomal plots of CNV alterations in 4 groups.

WES identifies mutation landscape and pathway enrichments in glioma

In the mutation landscape (Fig. 2a), of all 27 patients, 20 cases harboured MGMT promoter methylation and/or TERT promoter mutation. MUC4, ANKRD36C, LOC101060017, MUC16, RELN, TP53, CDC27, IDH1 and PCMTD1 were identified to be the most highly mutated genes in this cohort with the frequency more than 20% and a diverse range of mutations like missense, splice site and multiple hit mutation, which was highly compatible with previous glioma reports [22]. Subsequent gene ontology (GO) analysis was performed based on the mutated genes and the results showed that cellular biological processes such as positive regulation of catalytic activity, mRNA metabolic process and ectoderm development were usually altered in glioma (Fig. S2A). KEGG enrichment analysis uncovered frequent gene mutations in Ras, PI3K/Akt, FOXO, and ErbB signaling pathways (Fig. S2B). To investigate the crucial pathways induced by genetic alterations in LGG and HGG patients, we then compared the KEGG enrichment data in both groups and found that all the pathways involved in LGG were consistent with those in HGG (Fig. S2C, D), confirming that gene mutations, as well as their involved signaling pathways were not related to grades of glioma. When searching for the potential driver gene in glioma, we firstly ruled out those already well-known genes listed in other research and picked some candidates including SRPX, SIX2 and NOD1 because of relative high frequency (more than 11%) and conserved mutations (no more than 2 mutations in one gene). After the preliminary screening by the cell growth and wound healing assay, we innovatively found that SRPX mutations might be associated with the progression of glioma (data will show in Fig. 3), while mutations of SIX2 and NOD1 did not affect glioma cell growth and migration (Fig. S3A, B).

Fig. 2. SRPX might be one of the driver genes in glioma malignancy.

a Mutation landscape of glioma is profiled by WES. Top 60 mutated genes are listed in order of mutation frequency, and specific type of genetic mutations, MGMT methylation and TERT promoter mutation are also included to illustrate co-occurrence. SRPX is moderately mutated in this cohort (arrow hinted). b Mutation frequency of SRPX in LGG and HGG groups, WT (SRPX wild type), MUT (SRPX mutant). c Map of SRPX protein structure and the mutation sites. Functional domains of the altered proteins are based on UniProt database. d GO analysis of SRPX in this cohort. e mRNA levels of SRPX in normal and tumour tissue of glioma. RNA-Seq data were extracted from GTEX (normal) and TCGA (LGG and GBM) database. f Survival analysis of SRPX based on RNA-Seq and clinical data from TCGA (LGG and GBM) database.

Fig. 3. SRPX mutations are associated with aggressive tumour phenotype.

After transient transfection of vector, SRPX WT and mutant plasmids in LN229 and HS683 cells for 48 h, a cell viability assay, b wound healing assay, c colony formation assay, and d neurosphere formation assay were tested. Data were shown as mean ± SD at three independent experiments. ***, P < 0.001. **, P < 0.01. *, P < 0.05. VEC vector.

SRPX might be one of the driver genes in glioma malignancy

SRPX gene was moderately mutated (5/27, 19%) and the mutations consists of two in-frame deletions in our WES study: (i) p.L14DEL (1/27, 4%), specific alteration in nucleotide sequence GGCA > G; (ii) p.L23DEL (4/27, 15%), specific alteration in nucleotide sequence CGCA > C. Meanwhile, no other mutations such as point mutations, insertions or splice site mutations in the SRPX gene were observed in this cohort. SRPX mutations occurred in both LGG and HGG patients and the mutation rates in each group were 8% and 11%, respectively (Fig. 2b). SRPX is located on chromosome X and the protein structure contains five functional domains: initial signal peptide, sushi1, sushi2, HYR and sushi4. The two in-frame deletions caused a 14th or 23rd amino acid lysine deletion in signal peptide, which was proposed to interfere with signal transduction of SRPX (Fig. 2c). GO analysis based on our 27 glioma samples WES data revealed that SRPX was participated in cell proliferation, differentiation, migration and adhesion (Fig. 2d). In order to explore the general role of SRPX in glioma, we conducted integrated analysis of TCGA LGG and GBM exome studies and found that SRPX is highly expressed in tumour compared with normal tissue (Fig. 2e). Furthermore, patients harboured high level of SRPX showed a declined overall survival while patients with low expression of SRPX had a better prognosis (Fig. 2f). Therefore, it is promptly assumed that SRPX might be one of the driver genes in glioma malignancy.

SRPX mutations are associated with aggressive tumour phenotypes

To explore the function of SRPX and its in-frame mutations in glioma, we constructed plasmids of SRPX wild type (WT), SRPX p.L14 DEL and p.L23 DEL and determined the influence of WT and mutated SRPX on cell proliferation, migration, chemosensitivity to TMZ and stemness. Intriguingly we observed increased cell growth and wound healing of SRPX p.L14 DEL and p.L23 DEL mutants compared to SRPX WT in HS683 (low grade glioma) and LN229 (high grade glioma) cell lines (Fig. 3a, b). Colony formation assay further indicated that p.L14 DEL and p.L23 DEL mutants, compared with SRPX WT, exhibited much more colonies, and treatment with TMZ could notably reduce such acceleration, which suggested that SRPX mutations would be sensitive to TMZ to a certain extent (Fig. 3c). In addition, SRPX p.L14 DEL and p.L23 DEL mutants were also identified to facilitate neurosphere formation, with a remarkably larger size and higher numbers of neurosphere in LN229 than HS683 (Fig. 3d). Meanwhile, we also recognised that SRPX WT showed stronger ability of proliferation, migration, colony formation and stemness when compared with vector (Fig. 3). Later a promoting role of endogenous SRPX was confirmed in glioma because decreased cell growth and migration were observed after knockdown of SRPX (Fig. S4). All these results indicated that SRPX was associated with aggressive tumour phenotypes and mutations of SRPX furtherly exert encouraging functions in glioma.

SRPX p.L14 DEL and p.L23 DEL mutations enhance the transcriptional activation of AHR on SRPX promoter

As we identified SRPX mutations playing a facilitating role in glioma, next we would like to reveal the underlying mechanism. First, we detected the expression of SRPX in adherent and supernatant of glioma cells because both of p.L14 DEL and p.L23 DEL occurred in signal peptide in SRPX, and signal peptides were reported to carry information for protein secretion [23]. However, the result showed a higher expression of SRPX in p.L14 DEL and p.L23 DEL mutants compared to WT in adherent cells, but no difference was observed in supernatant (Fig. 4a), which hinted that SRPX mutation did not interfere with the protein secretion. Second, we confirmed a joint cytoplasm and nuclear localisation of SRPX through IF, and it showed that p.L14 DEL and p.L23 DEL mutations did not change the distribution of SRPX protein (Fig. 4b). Many studies demonstrated that somatic gene mutations are common in both oncogenes and tumour suppressors, as a result, they may either enhance or degrade the function of a protein including itself in multiple ways like regulation of protein degradation [24, 25], activation or inhibition of the gene transcription [26–28]. We then analysed the expression of SRPX under the treatment of MG-132, a common proteasome inhibitor, and found that MG-132 did not increase the stability of SRPX protein, suggesting that upregulation of SRPX protein level by p.L14 DEL and p.L23 DEL mutants was a proteasome-independent manner (Fig. 4c). Meanwhile we identified dramatically higher mRNA levels in SRPX p.L14 DEL and p.L23 DEL mutant group than SRPX WT (Fig. 4d), and Actinomycin D assay validated that p.L14 DEL and p.L23 DEL mutations did not influence mRNA stability (Fig. 4e). These results revealed that increased expression of SRPX induced by the two deletion mutations might be due to activation of gene transcription, but not post-transcriptional or translational regulation.

Fig. 4. SRPX p.L14 DEL and p.L23 DEL mutations enhance the transcriptional activation of AHR on SRPX promoter.

Cells in this figure (exclude in (f)) were all transiently transfected for 48 h with SRPX VEC/ WT/ p.L14DEL/ p.L23DEL plasmids, respectively. a Detection of SRPX protein expression in adherent LN229 and HS683 cells or supernatant by Western blotting. b Cellular localisation of SRPX showed by IF assay in LN229 cell. c Detection of SRPX protein expression in LN229 and HS683 cells with treatment of proteasome inhibitor MG-132 for 6 h. d Detection of SRPX mRNA level in LN229 cell. e Actinomycin D assay in LN229 cell. f Detection of SRPX protein expression after knockdown of AHR, CEBPA, YY1 and JUN in LN229 cell to estimate possible transcriptional factors of SRPX. g ChIP assay was performed in LN229 cells to confirm the binding of AHR to SRPX promoter. Promoter of SRPX was separated into 7 parts P1–P7. h Agarose gel electrophoresis of ChIP-PCR products in LN229 cells. i Detection of SRPX protein level after knockdown of AHR.

To elucidate the mechanism of SRPX p.L14 DEL and p.L23 DEL mutants’ promotion on gene transcription, we first screened the potential transcriptional factors of SRPX through combinations of four acknowledged TF database including PROMO, ChIP-Atlas, GTRD and hTFtarget. Four possible TFs (AHR, CEBPA, YY1, JUN) were selected (Fig. S5) and verified their impacts on SRPX expression in parent LN229 glioma cell. Notably, AHR was speculated to participate in increased transcription of SRPX as knockdown of AHR decreased SRPX protein expression (Fig. 4f). Furthermore, ChIP assay and ChIP-qPCR were performed in SRPX WT, p.L14 DEL and p.L23 DEL cells and result validated that AHR could bind to promoter of SRPX in Part 6, and p.L14 DEL and p.L23 DEL mutations evidently enhanced binding of AHR (Fig. 4g, h). Correspondingly, knockdown of AHR apparently reduced SRPX expression, which proved that p.L14 DEL and p.L23 DEL mutations promote transcription of itself through enhancing binding of AHR (Fig. 4i).

SRPX p.L14 DEL and p.L23 DEL mutations upregulate EGFR/Akt/ Nestin pathway in glioma

Since family member SRPX2 was reported to constitute complex with transmembrane receptors in cell signal transduction [29], we firstly tried to search for functional targets of SRPX in cell receptors. EGFR receptor and its downstream kinase Akt, as well as FOXO protein, known as key regulators in many cancers [30, 31], were exhibited with high mutation frequency in our previous KEGG enrichment analysis (Fig. S2B). Therefore, we investigated the status of EGFR/Akt and FOXO signaling pathways and stemness biomarkers including Nestin and CD133 to address the downstream of SRPX cascade. Specifically, upregulation of p-EGFR, p-Akt, and Nestin was observed in SRPX p.L14DEL and p.L23DEL mutants and no accordant change was identified in EGFR, Akt, CD133 and FOXO3A (Fig. 5a). Notably, knockdown of AHR apparently depleted such upregulation of p-EGFR, p-Akt, and Nestin in SRPX p. L14DEL and p. L23DEL mutants (Fig. 5b). These results showed that SRPX mutations regulate EGFR/Akt pathway rather than FOXO signaling pathway. Furthermore, we confirmed the interaction between SRPX and p-EGFR by IP experiments (Fig. 5c), and IF analysis also showed the co-localisation of SRPX and p-EGFR protein in cytoplasm and cell membrane (Fig. 5d), which hinted a ligand role of SRPX to EGFR receptor and then activation of EGFR pathway. Later, we proved that knockdown of SRPX in LN229 cells with p.L14DEL and p.L23DEL mutants obviously decreased activation of EGFR and Akt (Fig. 5e). Moreover, Akt inhibitor could abolish the enhancement of SRPX p.L14DEL and p.L23DEL mutations in glioma, as remarkable decrease of p-Akt and Nestin protein level was observed after the treatment of Akt inhibitor MK-2206 (Fig. 5f), as well as the capability in cell migration and stemness (Figs. 5g, h, S6A, B). These results uncovered a SRPX/ EGFR/ Akt/ Nestin axis and indicated a potential utilising of Akt inhibitor in glioma with SRPX mutations.

Fig. 5. SRPX p.L14 DEL and p.L23 DEL mutations upregulate EGFR/Akt/ Nestin pathway in glioma.

a Expression of downstream protein of p-EGFR, EGFR, p-Akt, Akt, Nestin, CD133 and FOXO3A after 48 h transfection of plasmid SRPX WT/ p.L14DEL/ p.L23DEL in LN229 and HS683 cells. b Detection of SRPX, p-EGFR, p-Akt and Nestin at protein expression after knockdown of AHR in LN229 cell. Cells were transfected with plasmid SRPX WT/ p. L14DEL/ p. L23DEL for 48 h before treatment of AHR siRNA. c IP assay to show the interaction of SRPX and p-EGFR after 48 h transfection of plasmid SRPX WT/ p.L14DEL/ p.L23DEL in LN229 cell. d IF assay to show co-localisation of SRPX and p-EGFR in LN229 and HS683 cells. e Detection of SRPX, p-EGFR, p-Akt and Nestin at protein expression level after knockdown of SRPX in LN229 cell. Cells were transfected with plasmid SRPX WT/ p.L14DEL/ p.L23DEL for 48 h before treatment of SRPX siRNA. f Regulation of Akt inhibitor MK-2206 on the expression of p-Akt and Nestin in SRPX WT/p.L14DEL/p.L23DEL stable cells (LN229). g Wound healing assay of SRPX WT/p.L14DEL/p.L23DEL stable cells (LN229) with 24 h treatment of Akt inhibitor. h Neurosphere formation assay of SRPX WT/p.L14DEL/p.L23DEL stable cells (LN229) with continuous treatment of Akt inhibitor no more than 9 days. Data were shown as mean ± SD at three independent experiments. **, P < 0.01. *, P < 0.05. NS.

SRPX mutations facilitate growth and progression of glioma in vivo

To verify the oncogenic function of SRPX mutations in vivo, we generated immortalised LN229 cell line of SRPX WT, p.L14 DEL and p.L23 DEL and injected the cells subcutaneously into nude mice. In contrast to SRPX WT, SRPX p.L14 DEL and p.L23 DEL mutants significantly facilitated tumour growth (Fig. 6a, b). Western blotting assay of mice xenograft tissue confirmed that SRPX p.L14 DEL and p.L23 DEL mutations upregulated SRPX expression, promoting activity of EGFR and Akt, and subsequently increasing Nestin level (Fig. 6c).

Fig. 6. SRPX mutations facilitate growth and progression of glioma in vivo.

a Tumour removed from each individual mouse of SRPX WT/p.L14DEL/p.L23DEL groups. b The tumour volume for SRPX WT/p.L14DEL/p.L23DEL LN229 stable cell lines injected mice. Statistical analysis was made among the groups. c Western blotting analysis of SRPX, p-EGFR, p-Akt and Nestin expression in representative mice xenograft tissues. d IHC of SRPX in human glioma patients who underwent WES in this cohort. Representative image of SRPX WT/p.L14DEL/p.L23DEL was shown on the left and quantification plot was made on the right. Low, Mid and High refers to expression of SRPX. e IHC of SRPX, p-EGFR, p-Akt and Nestin expression in human glioma patients who underwent WES in this cohort. Representative images of positive and negative signal were shown. f Pearson correlation analysis among SRPX and p-EGFR, p-Akt and Nestin in human glioma samples. IHC scores for different molecules were used to test the correlations. g A schematic model illustrates the potential mechanisms by which SRPX and p.L14DEL/p.L23DEL mutations promote the stemness and malignant progression of glioma.

Moreover, to confirm expression and subcellular localisation of SRPX WT and two in-frame deletion mutations and the translational information of the SRPX/ EGFR/ Akt/ Nestin axis in clinical significance, IHC staining was performed in samples of those glioma patients including 5 SRPX mutated and 22 non-mutated, who underwent WES analysis in this study. In accordance with the result of plasmid transfection in glioma cells, mutated SRPX was localised in cytoplasm and nucleus, and SRPX p.L14 DEL and p.L23 DEL mutated patients showed an essentially higher expression of SRPX, while SRPX WT patients displayed no or relatively low level of SRPX (Fig. 6d). Besides, specimens with high expression of SRPX were more likely to present with enhanced p-EGFR, p-Akt, and Nestin level, and a significantly positive correlation of SRPX with p-EGFR, p-Akt and Nestin were recognised by Pearson correlation analysis respectively (Fig. 6e, f).

Discussion

Recently 'omics' research has played a pioneering role in comprehensive understanding of tumour aetiology, pathogenesis, progression, as well as searching for therapeutic candidates, among of which WES and WGS are the most proven technology in glioma [32]. In this study, we undertook WES analysis of 27 Chinese glioma patients and obtained two innovative findings which could help to reveal evolution and provide new potential candidates in diagnosis and treatments of glioma. First, a novel driver gene SRPX with two mutations were found to facilitate stemness and progression of glioma through SRPX/ EGFR/ Akt/ Nestin pathway. Although our Chinese cohort revealed a moderate mutation frequency (19%) of SRPX, integrated analysis of TCGA White cohorts indicated a low mutation frequency less than 2% (Fig. S7). We assumed it might be related to a racial discrepancy because global data showed that malignant brain tumour incidence rates are higher in non-Hispanic White and lower in Asian or Pacific Islander individuals combined both sexes during 2013-2017 [33]. Besides, prior study also found some molecular racial disparities in glioma. For example, PTPN11 mutations tended to be more frequent in HGGs in African American patients than in Caucasian cases, while PTEN mutations and deleterious TP53 were less frequent in African American patients than in Caucasian cases [34]. This racial disparity might be a comprehensive consequence of genetic risk factors, ionising radiation, allergies, lifestyle and diet [35]. A large scale of glioma cohort study is required to explore the potential mechanisms behind this racial disparity in the future. Second, genomic profiles of glioma did not interfer with tumour grades since no significant distinction of mutation burden, mutated genes and involved signaling pathways were found in LGG and HGG patients. Although we tried to figure out the connection of mutation profiles to tumour relapse, we could not make any significant and generalisable conclusions because the limited sample size of recurrent glioma in this project. More WES data especially recurrent glioma patients, need to be collected in future study to enhance the reliability of the conclusions.

SRPX, confirmed to associate with some benign diseases like cerebral amyloid angiopathy and retinitis pigmentosa, is a little-known oncogene and here we elucidated the contributing role of SRPX in glioma. Two mutations of SRPX p.L14DEL and p.L23DEL were both validated to promote aggressive tumour phenotypes in glioma including cell proliferation, migration, stemness and tumour growth. And such aggressive behaviours were more evident in high grade glioma cell LN229 when compared with low grade glioma cell HS683, suggesting that more attentions to SRPX mutation status are recommended in HGG patients. Our findings enable additional driver genes considered in diagnosis and predicting prognosis of glioma.

To best of our knowledge, in-frame deletion mutations are common in both oncogenes and tumour suppressors, as a result, they may either enhance or eliminate the function of a protein including itself in multiple ways like activation or inactivation of the gene transcription and regulation of the protein degradation [24, 26, 28, 36, 37]. For instance, mutations of interferon regulatory factor 4 (IRF4) were usually detected in adult T-cell leukaemia and were transcriptionally activated downstream of T-cell receptor CD28 and NF-κB pathways, which induce the expansion of T lymphocytes and inhibit apoptosis [26]. Mutations in the FMS-like tyrosine kinase 3 (FLT3) gene in 13q12.2 increased cis interactions between the FLT3 promoter and another enhancer, subsequently upregulated allele-specific FLT3 expression, and led to ligand-independent activation of the receptor and downstream signaling [28]. Whereas mutated CALR induced megakaryocytic signaling to transform hematopoietic cells and underwent accelerated degradation and Golgi-mediated secretion in myeloproliferative neoplasms [37]. In our study, SRPX mutations were uncovered to upregulate SRPX protein expression through activation of gene transcription by promoting binding with transcriptional factor AHR. And these two mutations did not affect secretory function, cell distribution, protein degradation, or mRNA stability of SRPX.

Our results also elucidated the underlying mechanism of SRPX somatic mutations. EGFR/ Akt/ Nestin axis was identified to be downstream target of SRPX in glioma. Supported by findings in lung and pancreatic cancer, Akt could increase expression of Nestin and accelerate tumour stemness [38, 39]. Here we confirmed that SRPX mutations stimulated activity of EGFR and Akt, upregulated expression of stemness marker Nestin, and subsequently contributed to tumour growth, progression and therapy-resistance of glioma. Furthermore, application of Akt inhibitor might reduce such facilitation mediated by SRPX mutations, suggesting a new therapeutic potential of Akt inhibitor in glioma patients, especially those with SRPX mutations. Most of clinical trials for pan-Akt inhibitor MK-2206 are now in phase I or II studies. Recent researches revealed that monotherapy of MK-2206 in adenoid cystic carcinoma was failed [40]. But combination of oral MK-2206 135 mg per week with standard neoadjuvant chemotherapy could improve response in breast cancer patients. Adverse events related to MK-2206 included maculopapular rash, neutrophil decrease, hypertension and hyperglycaemia [41]. Incorporation of our findings, a combined therapy of MK-2206 with temozolomide would be recommended in SRPX WT or mutation with p.L14DEL and p.L23DEL patients, and overall survival and event-free survival should be evaluated as study assessments. Certainly, step-by-step clinical trials are further required to ensure efficient and safe utility of Akt inhibitor in glioma.

In conclusion, our WES data highlighted the landscape of all grades glioma in revealing a non-distinctive mutation profile and consistent signaling pathways between LGG and HGG gliomas. More importantly, we recognised a novel role of SRPX’s conserved mutations that p.L14 DEL and p.L23 DEL promote binding with transcriptional factor AHR, then upregulate SRPX expression and activate EGFR/ Akt/ Nestin axis, thus accelerating with stemness and malignant progression of glioma (Fig. 6g). We believe it demonstrating the potential targets in prosperous therapy of glioma.

Author contributions

LY, LS, and ST conceived and designed the study. ST and CQ performed experiment. MZ recruited subjects and provided methodology. ST, CQ, PZ, PX and ZZ analysed data. ST and LY wrote the manuscript. All authors approved the final manuscript.

Funding

This work was supported by the grants from the National Natural Science Foundation of China (81974466) and the Natural Science Foundation of Hunan Province, China (No: 2022JJ30059, 2023JJ40067).

Data availability

All data supporting the findings of this study are available from the lead corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The study was approved by the Medical Ethics Committee of Xiangya Hospital, Central South University. All clinical sample acquisition was carried out according to the institutional guidelines, and all subjects have signed informed consent. All animal studies were approved by the Institutional Animal Care and Use Committee of Central South University.

Supplementary information

The online version contains supplementary material available at 10.1038/s41416-025-03091-5.