Comparative proteomic analysis of astrocytoma tissues from patients with and without seizures

Abstract

Astrocytoma is a common type of glioma and a frequent cause of brain tumour-related epilepsy. Although the link between glioma and epilepsy is well established, the precise mechanisms underlying epileptogenesis in astrocytoma remain poorly understood. In this study, we performed proteomic analysis of astrocytoma tissue from patients with and without seizures using mass spectrometry-based techniques. We detected 131 differentially expressed proteins (42 upregulated and 89 downregulated). Proteins upregulated in patients with seizures were mostly related to an increase in energy metabolism. Moreover, glial fibrillary acidic protein, which is involved in maintaining normal axonal structures, was abnormally highly expressed in patients with seizures. Proteins downregulated in patients with seizures included those involved in trans-synaptic signalling and gamma-aminobutyric acid synaptic transmission. Interestingly, comparison of protein expression profiles from our cohort with those from a previous study of patients with epilepsy due to other causes showed that the collapsin response mediator protein family of axonal growth regulators was highly expressed only in patients with seizures due to astrocytomas. Further studies of the proteins identified here are required to determine their potential as biomarkers and therapeutic targets.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-87525-3.

Introduction

Brain tumour-related epilepsy accounts for approximately 12% of all cases of acquired epilepsy¹. Seizures are more prevalent in patients with gliomas, particularly low-grade gliomas², with a reported 60–90% of patients with this tumour type affected². Patients with gliomas who experience seizures are susceptible to complications such as cognitive deterioration, falls, and psychological stress. Most previous studies have focused on the prevalence of seizures and their risk factors in patients with all types of glioma^1–4. Although numerous risk factors have been identified, the reasons behind the seizures remain unknown, and more research is therefore needed to fully understand epileptogenesis and its underlying mechanisms. The methods of studying these mechanisms have been greatly improved by the development of new molecular technologies and omic laboratory techniques⁵.

Proteomic analysis of glioma-related epilepsy can detect pathway-specific changes by characterising tumours and the neural microenvironment in a high-throughput manner^6,7. This allows the identification of biomarkers to predict seizure risk, assess disease progression, and monitor treatment responses, thereby potentially improving patient diagnosis and prognosis^8,9. Few studies have investigated the molecular level, and more specifically, protein expression in patients with seizures^4,7,10. There is now compelling evidence that the presence of a glioma leads to complex epileptogenic changes governed by interactions between tumour cells and the local neural network^8,11.

Of the various types of gliomas, astrocytomas, which are characterised by complex neurobiological changes at different cellular/molecular network levels, are the most common, and frequently trigger epilepsy⁶. Although the link between gliomas and epilepsy is well established⁷, the precise mechanisms underlying epileptogenesis vary between different types of glioma. Astrocytomas have distinct protein expression profiles that determine their responses to seizures. Therefore, personalised strategies are required to address epilepsy in patients with different glioma types. A detailed evaluation of the molecular processes responsible for epileptogenesis in patients with astrocytomas is thus necessary to develop novel therapeutic strategies for this condition.

This study aimed to evaluate the differences in protein expression between patients with astrocytomas who do and do not experience seizures. We performed extensive proteomic profiling of these two groups of patients. Additionally, we investigated possible distinctions between seizures due to astrocytomas and those due to other causes.

Methods

Brain astrocytoma tissue samples were originally obtained from the Biobank of Songklanagarind Hospital between 1998 and 2023 for the large main research project. For this study, all tissue sample collected from 2020 to 2023 were selected based on their suitability for proteomic study, and consent was obtained for research purposes. The study included patients with astrocytoma who presented with or without seizures and underwent craniotomy and tumour resection. We recorded patient characteristics, including demographic data, presenting symptoms, neuroimaging findings, data on neurological function, management details including antiseizure medication use. Radiological features including tumour location, tumour size, tumour volume, and peritumoural oedema were assessed. Tumour volume was calculated based on the T2/FLAIR hyperintense areas. The diagnosis of epileptic seizures was based on the criteria established by the 2017 International League Against Epilepsy, and semiological seizure was classified preoperatively. The study was conducted in accordance with the ethical precepts laid down in the Declaration of Helsinki and approved by the Ethical Committee of the Faculty of Medicine, Prince of Songkla University (EC. 65-168-14-1). Informed consent was obtained from all the participants.

Sample preparation

Resected tumour tissues were snap-frozen and stored in liquid nitrogen until protein extraction. Frozen tissues were washed three times with phosphate-buffered saline and homogenised in radioimmunoprecipitation assay buffer (50 mM Tris, pH 7.5, 0.1% Nonidet P-40, 0.1% deoxycholate, 150 mM NaCl, and 4 mM EDTA) and a mixture of protease inhibitors (0.01% aprotinin, 10 mM sodium pyrophosphate, 2 mM sodium orthovanadate, and 1 mM phenylmethylsulfonyl fluoride). The total cellular lysate was centrifuged at 12,000 rpm for 20 min, and the supernatant was stored at − 80 °C until analysis. Protein concentrations in the cellular extracts were determined using bicinchoninic acid assays.

In-gel tryptic digestion

The extracted proteins (20 mg per sample) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis using stacking and resolving gel matrices composed of 4% and 12% acrylamide, respectively. The proteins were subjected to in-gel enzymatic digestion following the excision of 2 × 2 mm protein-containing gel portions. Subsequently, the proteins within the gel segments were denatured using dithiothreitol, while iodoacetamide was introduced to inhibit refolding. The denatured proteins were subjected to trypsin-mediated digestion, maintaining a trypsin-to-protein mass ratio of 1:50 (w/w), for 20 h at 37 °C.

The resulting peptides were purified and desalted using EMPORE™ C18 solid-phase extraction disks (3 M, Saint Paul, MN, USA), C18 beads, and the stage-tip technique¹². The purified peptide was dehydrated using a vacuum centrifugal concentrator (Labconco, Kansas City, MO, USA). The resulting dehydrated peptide was resuspended in 0.1% formic acid to a final volume of 20 µl, yielding a peptide concentration of 1 mg/ml.

Proteomic analysis

Protein identification and quantification were performed using liquid chromatography with tandem mass spectrometry. In brief, 2 µg peptide was injected into an ultra-high performance nanoflow LC (Eksigent Technologies LLC, Dublin, CA, USA) equipped with C18 trap and analytic columns (Nano Trap TP-1, 3 μm 120 Å, 10 mm × 0.075 mm and bioZen Peptide Polar C18 nanocolumn, 75 μm × 15 cm, C18 particle sizes of 3 μm, 120 Å; Phenomenex, Torrance, CA, USA) with gradient elution of 0.1% formic acid in water as solvent A and 0.1% formic acid in acetonitrile as solvent B over 105 min. The m/z data were acquired using a TripleTOF 6600+ system (AB Sciex LLC, Framingham, MA, USA) operated in positive ion mode on Sequential Window Acquisition of all Theoretical Mass Spectra (SWATH-MS). An information-dependent data acquisition (IDA) scan was performed at a 100 ppm precursor mass tolerance and a 0.2 Da fragment mass tolerance in the detection range 300–1800 Da. The data-independent acquisition (DIA) mode was operated in the mass window of the 7-m/z and 1-m/z overlapping window. All .wiff data were subjected to protein identification and quantification using Protein Pilot software version 5.0.2.0 (AB Sciex LLC) against the Swiss-Prot database (UniProtKB 2022_01) Homo sapiens, which includes 20,385 proteins. The search was performed using the following parameters: IDA carbaminomethyl fixed modification, trypsin-lysc digestion, one missed cleavage, monoisotopic mass, and < 0.01 false discovery rate (FDR); DIA 10-min extraction windows, 25 peptides/protein, 6 transitions/peptide, excluding shared peptides, 20 ppm XIC width, and < 0.01 FDR.

Statistical analysis

Protein intensity was normalised using multiple linear regression and log₂ transformation. Differential protein expression between the seizure and non-seizure groups was analysed with an empirical Bayesian t-test using the limma package in R¹³. The level of significant differential protein expression was set at fold change ≥ |1.5|, p ≤ 0.05. Differentially expressed proteins were further subjected to overrepresentation analysis using the clusterProfiler package in R¹⁴ and the Gene Ontology (GO), Kyoto Encyclopaedia of Genes and Genomes (KEGG), and Wikipathways databases. The Benjamini–Hochberg procedure was applied for p-value adjustment, if required. All statistical analyses and visualisations were performed using R software version 4.2.3 (The R Foundation for Statistical Computing, Vienna, Austria).

Obtaining data of epilepsy of non-cancer origins

To evaluate the proteomic profile of epilepsy of non-cancer origins, we obtained data from two independent cohorts. The first dataset obtained from Pires et al.¹⁵ comprised histological findings of formalin-fixed brain tissue obtained from autopsy epilepsy cases and non-epilepsy controls. Control cases comprised individuals who died from various intoxicating causes such as drowning, choking, and pulmonary embolism¹⁵. The second dataset obtained from Srivastava et al.¹⁶ consisted of histological findings of surgically resected brain tissue obtained from patients with mesial temporal lobe epilepsy and non-seizure autopsy controls. The differentially expressed proteins from all brain tissue regions in each cohort were combined, using statistical analysis criteria similar to those applied to our cohort.

Results

Clinical characteristics of patients with astrocytomas

Twenty patients with astrocytomas (10 with and 10 without seizures) were included in this study. The patients with and without seizures had mean ages of 46 and 45 years, respectively and tumour sizes of 138.46 and 127.6 cm³, respectively. Almost all patients with seizures (9/10) had presented with seizures. The clinical characteristics of the patients are summarised in Table 1.

Table 1.

Demographic and clinical characteristics of the patients.

Patient’s characteristics	Seizure (n = 10)	No seizure (n = 10)
Age—years (mean)	46.0	45.0
Female sex—no. (%)	3 (30.0)	7 (70.0)
Number of performed surgery
1	6	4
2	2	5
3	2	1
Tumour grading
Grade I	4	4
Grade II	4	5
Grade III	2	1
IDH-1 (-/+)	2/8	1/9
Location of tumour
Extratemporal	8	8
Temporal	2	2
Side of tumour
Right	5	6
Left	4	2
Bilateral	1	1
Seizure semiology
Focal onset with impaired awareness	2	0
Generalized seizure	6	0
Unknown onset	2	0
Seizure frequency
No in the past 3 month	3	0
1 or more but less than 1 per month	3	0
More than 1 per month but no more than 1 per week	3	0
More than 1 per week but not more than 1 per day every day	1	0
Adjuvant treatment
Radiotherapy	9	8
Chemotherapy	3	3
Other neurological symptoms
Headache	5	7
Weakness	4	4
Numbness	0	2
Cognitive impairment	1	3
Ataxia	0	2
Blurred vision	1	1
Asymptomatic	3	0
AEDs before surgery
Yes	10	0
No	0	5
Present with seizure
Yes	9	0
No	1	10
Post-surgical seizure
No	5	10
At least daily	2	0
At least weakly	0	0
At least months	3	0
Tumour volume (cm3)	138.46	127.61
Perilesional oedema
No	3	0
Mild	5	6
Moderate	2	1
Severe	0	3
Intra-tumoural haemorrhage
Yes	1	4
No	9	6
Last follow up (months)	45.5	31.8
MRS score
0	5	1
1	1	2
2	0	3
3	1	0
4	2	0
5	0	4
6	1	0
Status at last follow
Alive	9	10
Dead	1	0

SWATH-MS/MS proteomic analysis

Overall protein representation

SWATH-MS/MS on the entire cohort revealed 4113 distinct peptides corresponding to 537 unique proteins (Supplementary Table 1). The highest protein intensities (Supplementary Fig. 1, Supplementary Table 2A) were related to neuronal structure (GFAP, ACTB, F107A, MBP, and KCRB) and maintenance of the blood–brain barrier (ACTB and MBP). In addition, there was a noticeable intensity of albumin. The 15 lowest intensities proteins were related to enzymes (GANAB, NAKD2, ATPK, FUMH, INPP, CBR1, and PDXK) and transport (NAC2 and VP13B).

Differential protein expression between patients with and without seizures

Principal component analysis of the quantitative proteomic data revealed a distinct separation between seizure (red dots) and non-seizure (blue dots) group samples, with PC 1–3 variations of 27.40%, 13.42%, and 8.25%, respectively (Fig. 1A). Consequently, we performed differential expression analysis using the thresholds of 1.5-fold change in the seizure and non-seizure groups (p ≤ 0.05, FDR ≤ 0.2). We identified 131 proteins that were differentially expressed in patients with seizures compared to those without seizures (42 upregulated and 89 downregulated). The volcano plot of the highlighted proteins showed statistically significant changes in expression levels between astrocytomas with and without seizures (Fig. 1B, Supplementary Table 2B). The top 10 most upregulated and downregulated proteins were selected for a heatmap representation in Fig. 1C. Additionally, we performed correlation analysis of the differentially expressed proteins with the tumour volume. The result demonstrated that the proteins SPTB2, DPYL5, IGM, EFTU, and STX1B intensities correlated with tumour volume (Supplementary Fig. 2).

Fig. 1.

The general overview of proteomic expression in astrocytoma tissue. (A) Principal component analysis of the data distribution of astrocytoma with seizure and without seizure. (B) Volcano plot of differentially expressed protein between seizure and non-seizure astrocytoma tissue. Red: significant change between astrocytoma with seizure and without seizure, green and black: no significant change between astrocytoma with seizure and without seizure. (C) An intensity heatmap top 10 predominantly upregulated and downregulated proteins from astrocytoma tissue with and without seizure. Barplot indicates the tumor volume in cm³ in each sample.

Functional enrichment analysis of differentially expressed proteins

All significantly differentially expressed proteins were further analysed using GO and KEGG analyses. The GO enrichment analysis revealed several biological processes that were significantly upregulated in patients with seizures. The most enriched pathways were energy production-related pathways, including the nucleoside triphosphate family, tricarboxylic acid (TCA) metabolism, and NADH regeneration. Additionally, there was an observable increase in the expression of proteins related to protein localisation and transport, post-synaptic cytoskeleton organisation, hypoxic cellular responses, and neuronal cell adaptation (Fig. 2A, B). Furthermore, the KEGG pathway analysis revealed that several pathways were significantly enriched among the proteins upregulated in patients with seizures, such as carbon metabolism, amino acid biosynthesis, and glycolysis/gluconeogenesis. There was also an overrepresentation of pathways related to the TCA cycle and hypoxia-inducible factor (HIF)-1 signalling pathway; these findings were complementary to those of the GO analysis (Fig. 2C, D).

Fig. 2.

Functional enrichment analysis of differentially expressed proteins. (A) GO analysis of upregulate on astrocytoma patient with epilepsy. (B) Interaction network analysis of upregulate protein (from GO analysis). (C) KEGG analysis of upregulate on astrocytoma patient with epilepsy. (D) Interaction network analysis of downregulate protein (from KEGG analysis). (E) GO analysis of downregulate on astrocytoma patient with epilepsy. (F) Interaction network analysis of downregulate protein (from GO analysis).

Conversely, proteins related to normal neural development, cell stabilisation, and synaptic stability were downregulated in patients with seizures. The biological processes that were significantly downregulated in patients with seizures were regulation of trans-synaptic signalling, modulation of chemical synaptic transmission, negative regulation of cell projection organisation, gamma-aminobutyric acid (GABA)ergic synapses, and synaptic vesicle priming (Fig. 2E, F). However, no downregulated proteins were overrepresented in the seizure group according to the KEGG pathway analysis. All raw enrichment data are presented in Supplementary Table 2.

Comparison with epilepsy due to other causes

To further explore the differences in protein expression between patients with and without seizures, we compared our proteomic data with publicly available data on differential protein expression in patients with epilepsy of non-cancer origins. The result demonstrated that the protein dysregulation profile in patients with astrocytoma experiencing seizures exhibits distinct characteristics compared to that of patients from the other two cohorts (< 1% in our cohort vs. others and 6.96% between others), suggesting a unique molecular characteristics of tumour-associated epilepsy (Fig. 3A). GO analysis of these proteins demonstrated common pathways involving pyruvate and TCA metabolism and proteins related to neutrophil degranulation. While the proteins differentially expressed in Pires et al.’s cohort was related to energy metabolism, L1CAM interaction, and semaphorin interactions, those differentially expressed in the present study were related to autophagy and CRMPs families (Fig. 3B, C). The enrichment data tables are provided in Supplementary Table 3.

Fig. 3.

Comparisons of epileptic protein profiles between astrocytoma and seizures from non-cancer causes from Srivasta, et al. and Pires, et al. (A) Total number of proteins identified in each cohort. (B) Dotplot of enriched pathways according to their etiologies. (C) Interaction network analysis of the dysregulated pathways between seizures for astrocytoma and other causes.

Discussion

The presence of a glial tumour is a significant risk factor for epilepsy¹⁷. However, the exact reasons for this remain unclear. In this study, astrocytoma tumour proteomes from patients with and without epilepsy were analysed to identify abnormally expressed proteins that may cause abnormal neuroactivity. The protein with the highest expression was GFAP, an intermediate filament protein found only in astrocytes^11,18. High levels of GFAP have also been observed in the blood of patients with epilepsy^19,20. The abnormally high intensity found in the present study may be due to increased GFAP expression in the tumour itself, as well as increased transcription of GFAP mRNA as a result of long-term epilepsy²¹. Accordingly, inhibiting GFAP activity may represent a potential therapeutic strategy as it may simultaneously target tumour and seizure activity²².

Other highly expressed proteins were mainly related to neuronal structure and maintenance of the blood–brain barrier. These findings are consistent with those of previous studies on astrocytomas²³. Our finding of high albumin levels is also consistent with the results of a study by Hashemi et al., which reported upregulation of albumin in astrocytomas compared to normal tissues²⁴. Although glial cells can synthesise albumin²⁵, the high levels of albumin detected in the present study probably indicates poor maintenance of the blood–brain barrier by astrocytomas, rather than increased synthesis^26,27. Notably, a significant upregulation in IDHP protein was observed in patients with epilepsy. DHP is a member of the IDH2 gene family, whose dysregulation is associated with a heightened incidence of preoperative seizures in patients with glioma, likely attributable to disruptions within the TCA cycle²⁸. Conversely, less highly expressed proteins were mostly enzymes, which do not require high levels of expression to function; further studies on protein phosphorylation could provide more information on their functional activity.

Proteins that were upregulated in patients with seizures compared to patients without seizures were mostly related to an increase in energy metabolism, particularly purine and adenosine derivatives and TCA metabolism, both of which are linked to neuronal hyperexcitability²⁹. Additionally, enrichment in the cellular response to hypoxia pathway, particularly HIF-1 signalling, indicated that oxidative stress within the hypoxic tumour microenvironment contributes to enhanced ATP production³⁰. Excessive ATP production can result in high purine metabolite levels that are detectable in the blood of patients with epilepsy³¹.

Conversely, proteins that were downregulated in patients with epilepsy were predominantly involved in the regulation of trans-synaptic signalling and synaptic transmission of GABA, the principal inhibitory neurotransmitter in the brain³². This could be a result of altered GABA signalling caused by glutamate, which is released by astrocytomas to facilitate tumour invasion^10,17. Additionally, the complement inhibitor CD59 was downregulated in patients with seizures. The loss of CD59 expression has been shown to impair GABAergic synaptic transmission by disrupting Ca²⁺ synaptic and SNARE complex assembly³³. This suggests that targeted enhancement of GABAergic signaling might be more effective than general anticonvulsants in these patients.

To assess the difference in epileptogenesis between patients with and without cancer, we compared our proteomic data with those of a study including patients with epilepsy of non-cancer origins¹⁵. Interestingly, the protein expression profiles differed widely, with only the pyruvate and TCA cycle pathways. However, there were a common pathway in neuronal structural molecules L1 and semaphorins between our study and the studies by Pires et al.^34,35. These shared pathways suggest a possible loss of neuronal structure in epilepsy due to cancer, as well as in epilepsy from hypoxic events, which is unique to epilepsy from natural disease as presented in Srivastava et al.’s cohort. Conversely, unique proteins dysregulated in patients with astrocytomas were involved in autophagy and amino acid metabolism; an abnormally high expression of CRMP proteins was also observed. CRMP family members regulate axonal growth and regeneration³⁶; however, abnormal expression of these proteins can result in tumour cell metastasis and invasion³⁷. Additionally, as autoantigens, CRMP proteins are responsible for various neurological paraneoplastic disorders³⁶. Further functional studies of this pathway could potentially elucidate the aetiopathophysiology of seizures in patients with brain cancer.

This study has certain limitations. First, although all tumours originated from patients of comparable ages, there remained some heterogeneity in terms of seizure semiology and surgical treatment. Second, the study only examined the expression of proteins, not their functional activity. Further analysis of brain phosphoproteomics could potentially identify highly activated proteins that are the primary drivers of epileptogenesis. Finally, the relatively small sample size means that further validation of the identified proteins is required before they can be considered for use as biomarkers.

In conclusion, our proteomic analysis revealed distinct molecular signatures in astrocytoma patients with epilepsy, mainly by upregulation of energy metabolism pathways and downregulation of inhibitory neurotransmission, suggesting potential therapeutic value in GABAergic enhancement drug. Moreover, we found that abnormal pyruvate and TCA metabolism are commonly present in patients with epilepsy. However, the CMRP-related pathways represent the most likely pathophysiological cause of epilepsy only in patients with astrocytoma. Future studies focusing on protein phosphorylation and functional pathway analysis could further elucidate more dynamic pathophysiology of seizures activity in patients with brain cancer.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Kanitpong Phabphal: designed, prepared and conducted study, writing-original draft, writing-review & editing. Thanakorn Khaosuwan, Kittinun Leetanaporn: prepared, evaluation and analyzed, writing-original draft, writing-review & editing. Raphatphorn Navakanitworakul: writing-review & editing. Anukoon Kaewborisutsakul, Thara Thuntanatip: prepared the study. Pongsakorn Choochuen: conducted testing. Surasak Sangkhathat: designed and prepared the study. Wararat Chiangjong: conducted testing.

Data availability

The search result and All MS/MS .wiff data are deposited in the ProteomeXchange Consortium http://proteomecentral.proteomexchange.org via jPOST partner https://repository.jpostdb.org/entry/JPST003365.0). The dataset identifiers are JPST003365 and PXD055963.

Declarations

Competing interests

The authors declare no competing interests.